Hugging Face's logo

rizla
/
rizla-156

PEFT
TensorBoard
Safetensors
mixtral
Generated from Trainer
4-bit precision
bitsandbytes
Model card Files Metrics Community
rizla commited on
Commit
8107860
·
verified ·
1 Parent(s): 40fd083

Upload folder using huggingface_hub

Browse files
Files changed (28) hide show
  1. README.md +154 -0
  2. adapter_config.json +33 -0
  3. adapter_model.bin +3 -0
  4. checkpoint-601/README.md +204 -0
  5. checkpoint-601/adapter_config.json +33 -0
  6. checkpoint-601/adapter_model.safetensors +3 -0
  7. checkpoint-601/global_step601/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt +3 -0
  8. checkpoint-601/global_step601/zero_pp_rank_0_mp_rank_00_model_states.pt +3 -0
  9. checkpoint-601/latest +1 -0
  10. checkpoint-601/rng_state.pth +3 -0
  11. checkpoint-601/scheduler.pt +3 -0
  12. checkpoint-601/trainer_state.json +3627 -0
  13. checkpoint-601/training_args.bin +3 -0
  14. checkpoint-601/zero_to_fp32.py +592 -0
  15. config.json +46 -0
  16. runs/Feb03_03-36-39_19503a009d2c/events.out.tfevents.1706931401.19503a009d2c.4586.0 +3 -0
  17. runs/Feb03_03-38-23_19503a009d2c/events.out.tfevents.1706931504.19503a009d2c.4673.0 +3 -0
  18. runs/Feb03_03-50-56_19503a009d2c/events.out.tfevents.1706932258.19503a009d2c.5617.0 +3 -0
  19. runs/Feb03_03-53-37_19503a009d2c/events.out.tfevents.1706932424.19503a009d2c.5964.0 +3 -0
  20. runs/Feb03_03-55-24_19503a009d2c/events.out.tfevents.1706932529.19503a009d2c.6106.0 +3 -0
  21. runs/Feb03_03-58-43_19503a009d2c/events.out.tfevents.1706932728.19503a009d2c.6196.0 +3 -0
  22. runs/Feb03_04-04-38_19503a009d2c/events.out.tfevents.1706933083.19503a009d2c.6301.0 +3 -0
  23. runs/Feb03_04-07-08_19503a009d2c/events.out.tfevents.1706933232.19503a009d2c.6407.0 +3 -0
  24. runs/Feb03_04-11-41_19503a009d2c/events.out.tfevents.1706933506.19503a009d2c.6498.0 +3 -0
  25. runs/Feb03_04-13-40_19503a009d2c/events.out.tfevents.1706933624.19503a009d2c.6813.0 +3 -0
  26. special_tokens_map.json +35 -0
  27. tokenizer.model +3 -0
  28. tokenizer_config.json +52 -0
README.md CHANGED
@@ -1,3 +1,157 @@
1
  ---
2
  license: cc-by-nc-nd-4.0
 
 
 
 
 
 
 
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  license: cc-by-nc-nd-4.0
3
+ library_name: peft
4
+ tags:
5
+ - generated_from_trainer
6
+ base_model: rizla/rizla-17
7
+ model-index:
8
+ - name: lorazapam-out
9
+ results: []
10
  ---
11
+
12
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
13
+ should probably proofread and complete it, then remove this comment. -->
14
+
15
+ [<img src="https://raw.githubusercontent.com/OpenAccess-AI-Collective/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/OpenAccess-AI-Collective/axolotl)
16
+ <details><summary>See axolotl config</summary>
17
+
18
+ axolotl version: `0.4.0`
19
+ ```yaml
20
+ base_model: rizla/rizla-17
21
+ model_type: AutoModelForCausalLM
22
+ tokenizer_type: LlamaTokenizer
23
+ trust_remote_code: true
24
+ load_in_8bit: false
25
+ load_in_4bit: true
26
+ strict: false
27
+
28
+ datasets:
29
+ - path: meta-math/MetaMathQA-40K
30
+ type:
31
+ system_prompt: "You are an expert problem solver who is great at teaching how to solve problems via first principles reasoning"
32
+ field_system: system
33
+ field_instruction: query
34
+ field_output: response
35
+ format: "[INST] {instruction} [/INST]"
36
+ no_input_format: "[INST] {instruction} [/INST]"
37
+
38
+ dataset_prepared_path: last_run_prepared
39
+ val_set_size: 0.0
40
+ output_dir: ./lorazapam-out
41
+ ## You can optionally freeze the entire model and unfreeze a subset of parameters
42
+ # - lm_head.*
43
+ # - model.embed_tokens.*
44
+ # - model.layers.2[0-9]+.block_sparse_moe.gate.*
45
+ # - model.layers.2[0-9]+.block_sparse_moe.experts.*
46
+ # - model.layers.3[0-9]+.block_sparse_moe.gate.*
47
+ # - model.layers.3[0-9]+.block_sparse_moe.experts.*
48
+
49
+ model_config:
50
+ output_router_logits: true
51
+
52
+ adapter: qlora
53
+ lora_model_dir:
54
+
55
+ sequence_len: 512
56
+ sample_packing: true
57
+ pad_to_sequence_len: true
58
+
59
+ lora_r: 32
60
+ lora_alpha: 16
61
+ lora_dropout: 0.05
62
+ lora_target_linear: true
63
+ lora_fan_in_fan_out:
64
+
65
+ wandb_project:
66
+ wandb_entity:
67
+ wandb_watch:
68
+ wandb_name:
69
+ wandb_log_model:
70
+
71
+ gradient_accumulation_steps: 2
72
+ micro_batch_size: 16
73
+ num_epochs: 1
74
+ optimizer: adamw_bnb_8bit
75
+ lr_scheduler: cosine
76
+ learning_rate: 0.0002
77
+
78
+ train_on_inputs: false
79
+ group_by_length: false
80
+ bf16: true
81
+ fp16:
82
+ tf32: false
83
+
84
+ gradient_checkpointing: true
85
+ early_stopping_patience:
86
+ resume_from_checkpoint:
87
+ local_rank:
88
+ logging_steps: 1
89
+ xformers_attention: false
90
+ flash_attention: true
91
+
92
+ loss_watchdog_threshold: 5.0
93
+ loss_watchdog_patience: 3
94
+
95
+ warmup_steps: 10
96
+ evals_per_epoch: 1
97
+ eval_table_size:
98
+ eval_table_max_new_tokens: 128
99
+ saves_per_epoch: 1
100
+ debug:
101
+ # deepspeed: deepspeed_configs/zero_1.json
102
+ weight_decay: 0.0
103
+ fsdp:
104
+ fsdp_config:
105
+ special_tokens:
106
+ bos_token: "<s>"
107
+ eos_token: "</s>"
108
+ unk_token: "<unk>"
109
+
110
+ ```
111
+
112
+ </details><br>
113
+
114
+ # lorazapam-out
115
+
116
+ This model is a fine-tuned version of [rizla/rizla-17](https://huggingface.co/rizla/rizla-17) on the None dataset.
117
+
118
+ ## Model description
119
+
120
+ More information needed
121
+
122
+ ## Intended uses & limitations
123
+
124
+ More information needed
125
+
126
+ ## Training and evaluation data
127
+
128
+ More information needed
129
+
130
+ ## Training procedure
131
+
132
+ ### Training hyperparameters
133
+
134
+ The following hyperparameters were used during training:
135
+ - learning_rate: 0.0002
136
+ - train_batch_size: 16
137
+ - eval_batch_size: 16
138
+ - seed: 42
139
+ - distributed_type: multi-GPU
140
+ - gradient_accumulation_steps: 2
141
+ - total_train_batch_size: 32
142
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
143
+ - lr_scheduler_type: cosine
144
+ - lr_scheduler_warmup_steps: 10
145
+ - num_epochs: 1
146
+
147
+ ### Training results
148
+
149
+
150
+
151
+ ### Framework versions
152
+
153
+ - PEFT 0.8.2
154
+ - Transformers 4.38.0.dev0
155
+ - Pytorch 2.1.2+cu121
156
+ - Datasets 2.16.1
157
+ - Tokenizers 0.15.0
adapter_config.json ADDED
@@ -0,0 +1,33 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "rizla/rizla-17",
5
+ "bias": "none",
6
+ "fan_in_fan_out": null,
7
+ "inference_mode": true,
8
+ "init_lora_weights": true,
9
+ "layers_pattern": null,
10
+ "layers_to_transform": null,
11
+ "loftq_config": {},
12
+ "lora_alpha": 16,
13
+ "lora_dropout": 0.05,
14
+ "megatron_config": null,
15
+ "megatron_core": "megatron.core",
16
+ "modules_to_save": null,
17
+ "peft_type": "LORA",
18
+ "r": 32,
19
+ "rank_pattern": {},
20
+ "revision": null,
21
+ "target_modules": [
22
+ "v_proj",
23
+ "w2",
24
+ "w1",
25
+ "k_proj",
26
+ "gate",
27
+ "o_proj",
28
+ "w3",
29
+ "q_proj"
30
+ ],
31
+ "task_type": "CAUSAL_LM",
32
+ "use_rslora": false
33
+ }
adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b69b55ff756739b2cfbb33842728f2fabba87d84afac92381411bc636db6471a
3
+ size 352960418
checkpoint-601/README.md ADDED
@@ -0,0 +1,204 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ base_model: rizla/rizla-17
4
+ ---
5
+
6
+ # Model Card for Model ID
7
+
8
+ <!-- Provide a quick summary of what the model is/does. -->
9
+
10
+
11
+
12
+ ## Model Details
13
+
14
+ ### Model Description
15
+
16
+ <!-- Provide a longer summary of what this model is. -->
17
+
18
+
19
+
20
+ - **Developed by:** [More Information Needed]
21
+ - **Funded by [optional]:** [More Information Needed]
22
+ - **Shared by [optional]:** [More Information Needed]
23
+ - **Model type:** [More Information Needed]
24
+ - **Language(s) (NLP):** [More Information Needed]
25
+ - **License:** [More Information Needed]
26
+ - **Finetuned from model [optional]:** [More Information Needed]
27
+
28
+ ### Model Sources [optional]
29
+
30
+ <!-- Provide the basic links for the model. -->
31
+
32
+ - **Repository:** [More Information Needed]
33
+ - **Paper [optional]:** [More Information Needed]
34
+ - **Demo [optional]:** [More Information Needed]
35
+
36
+ ## Uses
37
+
38
+ <!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
39
+
40
+ ### Direct Use
41
+
42
+ <!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
43
+
44
+ [More Information Needed]
45
+
46
+ ### Downstream Use [optional]
47
+
48
+ <!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
49
+
50
+ [More Information Needed]
51
+
52
+ ### Out-of-Scope Use
53
+
54
+ <!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
55
+
56
+ [More Information Needed]
57
+
58
+ ## Bias, Risks, and Limitations
59
+
60
+ <!-- This section is meant to convey both technical and sociotechnical limitations. -->
61
+
62
+ [More Information Needed]
63
+
64
+ ### Recommendations
65
+
66
+ <!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
67
+
68
+ Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
69
+
70
+ ## How to Get Started with the Model
71
+
72
+ Use the code below to get started with the model.
73
+
74
+ [More Information Needed]
75
+
76
+ ## Training Details
77
+
78
+ ### Training Data
79
+
80
+ <!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
81
+
82
+ [More Information Needed]
83
+
84
+ ### Training Procedure
85
+
86
+ <!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
87
+
88
+ #### Preprocessing [optional]
89
+
90
+ [More Information Needed]
91
+
92
+
93
+ #### Training Hyperparameters
94
+
95
+ - **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
96
+
97
+ #### Speeds, Sizes, Times [optional]
98
+
99
+ <!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
100
+
101
+ [More Information Needed]
102
+
103
+ ## Evaluation
104
+
105
+ <!-- This section describes the evaluation protocols and provides the results. -->
106
+
107
+ ### Testing Data, Factors & Metrics
108
+
109
+ #### Testing Data
110
+
111
+ <!-- This should link to a Dataset Card if possible. -->
112
+
113
+ [More Information Needed]
114
+
115
+ #### Factors
116
+
117
+ <!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
118
+
119
+ [More Information Needed]
120
+
121
+ #### Metrics
122
+
123
+ <!-- These are the evaluation metrics being used, ideally with a description of why. -->
124
+
125
+ [More Information Needed]
126
+
127
+ ### Results
128
+
129
+ [More Information Needed]
130
+
131
+ #### Summary
132
+
133
+
134
+
135
+ ## Model Examination [optional]
136
+
137
+ <!-- Relevant interpretability work for the model goes here -->
138
+
139
+ [More Information Needed]
140
+
141
+ ## Environmental Impact
142
+
143
+ <!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
144
+
145
+ Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
146
+
147
+ - **Hardware Type:** [More Information Needed]
148
+ - **Hours used:** [More Information Needed]
149
+ - **Cloud Provider:** [More Information Needed]
150
+ - **Compute Region:** [More Information Needed]
151
+ - **Carbon Emitted:** [More Information Needed]
152
+
153
+ ## Technical Specifications [optional]
154
+
155
+ ### Model Architecture and Objective
156
+
157
+ [More Information Needed]
158
+
159
+ ### Compute Infrastructure
160
+
161
+ [More Information Needed]
162
+
163
+ #### Hardware
164
+
165
+ [More Information Needed]
166
+
167
+ #### Software
168
+
169
+ [More Information Needed]
170
+
171
+ ## Citation [optional]
172
+
173
+ <!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
174
+
175
+ **BibTeX:**
176
+
177
+ [More Information Needed]
178
+
179
+ **APA:**
180
+
181
+ [More Information Needed]
182
+
183
+ ## Glossary [optional]
184
+
185
+ <!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
186
+
187
+ [More Information Needed]
188
+
189
+ ## More Information [optional]
190
+
191
+ [More Information Needed]
192
+
193
+ ## Model Card Authors [optional]
194
+
195
+ [More Information Needed]
196
+
197
+ ## Model Card Contact
198
+
199
+ [More Information Needed]
200
+
201
+
202
+ ### Framework versions
203
+
204
+ - PEFT 0.8.2
checkpoint-601/adapter_config.json ADDED
@@ -0,0 +1,33 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "rizla/rizla-17",
5
+ "bias": "none",
6
+ "fan_in_fan_out": null,
7
+ "inference_mode": true,
8
+ "init_lora_weights": true,
9
+ "layers_pattern": null,
10
+ "layers_to_transform": null,
11
+ "loftq_config": {},
12
+ "lora_alpha": 16,
13
+ "lora_dropout": 0.05,
14
+ "megatron_config": null,
15
+ "megatron_core": "megatron.core",
16
+ "modules_to_save": null,
17
+ "peft_type": "LORA",
18
+ "r": 32,
19
+ "rank_pattern": {},
20
+ "revision": null,
21
+ "target_modules": [
22
+ "v_proj",
23
+ "w2",
24
+ "w1",
25
+ "k_proj",
26
+ "gate",
27
+ "o_proj",
28
+ "w3",
29
+ "q_proj"
30
+ ],
31
+ "task_type": "CAUSAL_LM",
32
+ "use_rslora": false
33
+ }
checkpoint-601/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e44ce263e6fd885f50d82ca515b9325375b43ee36ededb75acf161ce88bc2e41
3
+ size 48
checkpoint-601/global_step601/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0efcfd30ef0e98aaef40ec83c39cd06ede030aa4a9addc567cd9df1c8c4cb175
3
+ size 1058858128
checkpoint-601/global_step601/zero_pp_rank_0_mp_rank_00_model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9b4831e463f5f6d0efc2c9322e9e25477305bae9dfc55581c3cb8f6711a0512d
3
+ size 246308720
checkpoint-601/latest ADDED
@@ -0,0 +1 @@
 
 
1
+ global_step601
checkpoint-601/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:79c6866b8b8da812d24116d346cafad093a94c83cbc28645c27f1f56139f31e7
3
+ size 14244
checkpoint-601/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d0719fbae2fb6995577344ae4fe1584ae2a14167fbf1694858b0016649c911ae
3
+ size 1064
checkpoint-601/trainer_state.json ADDED
@@ -0,0 +1,3627 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 1.0,
5
+ "eval_steps": 500,
6
+ "global_step": 601,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.0,
13
+ "learning_rate": 2e-05,
14
+ "loss": 0.547,
15
+ "step": 1
16
+ },
17
+ {
18
+ "epoch": 0.0,
19
+ "learning_rate": 4e-05,
20
+ "loss": 0.5148,
21
+ "step": 2
22
+ },
23
+ {
24
+ "epoch": 0.0,
25
+ "learning_rate": 6e-05,
26
+ "loss": 0.5241,
27
+ "step": 3
28
+ },
29
+ {
30
+ "epoch": 0.01,
31
+ "learning_rate": 8e-05,
32
+ "loss": 0.3872,
33
+ "step": 4
34
+ },
35
+ {
36
+ "epoch": 0.01,
37
+ "learning_rate": 0.0001,
38
+ "loss": 0.3484,
39
+ "step": 5
40
+ },
41
+ {
42
+ "epoch": 0.01,
43
+ "learning_rate": 0.00012,
44
+ "loss": 0.2567,
45
+ "step": 6
46
+ },
47
+ {
48
+ "epoch": 0.01,
49
+ "learning_rate": 0.00014,
50
+ "loss": 0.2197,
51
+ "step": 7
52
+ },
53
+ {
54
+ "epoch": 0.01,
55
+ "learning_rate": 0.00016,
56
+ "loss": 0.2134,
57
+ "step": 8
58
+ },
59
+ {
60
+ "epoch": 0.01,
61
+ "learning_rate": 0.00018,
62
+ "loss": 0.188,
63
+ "step": 9
64
+ },
65
+ {
66
+ "epoch": 0.02,
67
+ "learning_rate": 0.0002,
68
+ "loss": 0.2067,
69
+ "step": 10
70
+ },
71
+ {
72
+ "epoch": 0.02,
73
+ "learning_rate": 0.00019999858715745195,
74
+ "loss": 0.1878,
75
+ "step": 11
76
+ },
77
+ {
78
+ "epoch": 0.02,
79
+ "learning_rate": 0.00019999434866973016,
80
+ "loss": 0.1898,
81
+ "step": 12
82
+ },
83
+ {
84
+ "epoch": 0.02,
85
+ "learning_rate": 0.00019998728465660105,
86
+ "loss": 0.1904,
87
+ "step": 13
88
+ },
89
+ {
90
+ "epoch": 0.02,
91
+ "learning_rate": 0.0001999773953176713,
92
+ "loss": 0.157,
93
+ "step": 14
94
+ },
95
+ {
96
+ "epoch": 0.02,
97
+ "learning_rate": 0.00019996468093238257,
98
+ "loss": 0.1442,
99
+ "step": 15
100
+ },
101
+ {
102
+ "epoch": 0.03,
103
+ "learning_rate": 0.00019994914186000328,
104
+ "loss": 0.1884,
105
+ "step": 16
106
+ },
107
+ {
108
+ "epoch": 0.03,
109
+ "learning_rate": 0.00019993077853961872,
110
+ "loss": 0.1518,
111
+ "step": 17
112
+ },
113
+ {
114
+ "epoch": 0.03,
115
+ "learning_rate": 0.00019990959149011848,
116
+ "loss": 0.1615,
117
+ "step": 18
118
+ },
119
+ {
120
+ "epoch": 0.03,
121
+ "learning_rate": 0.00019988558131018186,
122
+ "loss": 0.1421,
123
+ "step": 19
124
+ },
125
+ {
126
+ "epoch": 0.03,
127
+ "learning_rate": 0.00019985874867826096,
128
+ "loss": 0.1438,
129
+ "step": 20
130
+ },
131
+ {
132
+ "epoch": 0.03,
133
+ "learning_rate": 0.00019982909435256144,
134
+ "loss": 0.1365,
135
+ "step": 21
136
+ },
137
+ {
138
+ "epoch": 0.04,
139
+ "learning_rate": 0.00019979661917102115,
140
+ "loss": 0.127,
141
+ "step": 22
142
+ },
143
+ {
144
+ "epoch": 0.04,
145
+ "learning_rate": 0.00019976132405128647,
146
+ "loss": 0.145,
147
+ "step": 23
148
+ },
149
+ {
150
+ "epoch": 0.04,
151
+ "learning_rate": 0.00019972320999068636,
152
+ "loss": 0.1647,
153
+ "step": 24
154
+ },
155
+ {
156
+ "epoch": 0.04,
157
+ "learning_rate": 0.0001996822780662041,
158
+ "loss": 0.137,
159
+ "step": 25
160
+ },
161
+ {
162
+ "epoch": 0.04,
163
+ "learning_rate": 0.00019963852943444702,
164
+ "loss": 0.1427,
165
+ "step": 26
166
+ },
167
+ {
168
+ "epoch": 0.04,
169
+ "learning_rate": 0.0001995919653316137,
170
+ "loss": 0.1156,
171
+ "step": 27
172
+ },
173
+ {
174
+ "epoch": 0.05,
175
+ "learning_rate": 0.000199542587073459,
176
+ "loss": 0.1295,
177
+ "step": 28
178
+ },
179
+ {
180
+ "epoch": 0.05,
181
+ "learning_rate": 0.00019949039605525703,
182
+ "loss": 0.1465,
183
+ "step": 29
184
+ },
185
+ {
186
+ "epoch": 0.05,
187
+ "learning_rate": 0.00019943539375176164,
188
+ "loss": 0.1371,
189
+ "step": 30
190
+ },
191
+ {
192
+ "epoch": 0.05,
193
+ "learning_rate": 0.00019937758171716468,
194
+ "loss": 0.1276,
195
+ "step": 31
196
+ },
197
+ {
198
+ "epoch": 0.05,
199
+ "learning_rate": 0.00019931696158505223,
200
+ "loss": 0.1166,
201
+ "step": 32
202
+ },
203
+ {
204
+ "epoch": 0.05,
205
+ "learning_rate": 0.00019925353506835826,
206
+ "loss": 0.1194,
207
+ "step": 33
208
+ },
209
+ {
210
+ "epoch": 0.06,
211
+ "learning_rate": 0.00019918730395931649,
212
+ "loss": 0.1146,
213
+ "step": 34
214
+ },
215
+ {
216
+ "epoch": 0.06,
217
+ "learning_rate": 0.00019911827012940946,
218
+ "loss": 0.1539,
219
+ "step": 35
220
+ },
221
+ {
222
+ "epoch": 0.06,
223
+ "learning_rate": 0.0001990464355293158,
224
+ "loss": 1.9511,
225
+ "step": 36
226
+ },
227
+ {
228
+ "epoch": 0.06,
229
+ "learning_rate": 0.00019897180218885507,
230
+ "loss": 0.1222,
231
+ "step": 37
232
+ },
233
+ {
234
+ "epoch": 0.06,
235
+ "learning_rate": 0.00019889437221693053,
236
+ "loss": 0.1223,
237
+ "step": 38
238
+ },
239
+ {
240
+ "epoch": 0.06,
241
+ "learning_rate": 0.0001988141478014693,
242
+ "loss": 0.1353,
243
+ "step": 39
244
+ },
245
+ {
246
+ "epoch": 0.07,
247
+ "learning_rate": 0.00019873113120936074,
248
+ "loss": 0.1265,
249
+ "step": 40
250
+ },
251
+ {
252
+ "epoch": 0.07,
253
+ "learning_rate": 0.00019864532478639234,
254
+ "loss": 0.1288,
255
+ "step": 41
256
+ },
257
+ {
258
+ "epoch": 0.07,
259
+ "learning_rate": 0.00019855673095718336,
260
+ "loss": 0.128,
261
+ "step": 42
262
+ },
263
+ {
264
+ "epoch": 0.07,
265
+ "learning_rate": 0.0001984653522251165,
266
+ "loss": 0.1297,
267
+ "step": 43
268
+ },
269
+ {
270
+ "epoch": 0.07,
271
+ "learning_rate": 0.00019837119117226688,
272
+ "loss": 0.1265,
273
+ "step": 44
274
+ },
275
+ {
276
+ "epoch": 0.07,
277
+ "learning_rate": 0.0001982742504593294,
278
+ "loss": 0.1269,
279
+ "step": 45
280
+ },
281
+ {
282
+ "epoch": 0.08,
283
+ "learning_rate": 0.00019817453282554333,
284
+ "loss": 0.1407,
285
+ "step": 46
286
+ },
287
+ {
288
+ "epoch": 0.08,
289
+ "learning_rate": 0.00019807204108861502,
290
+ "loss": 0.1311,
291
+ "step": 47
292
+ },
293
+ {
294
+ "epoch": 0.08,
295
+ "learning_rate": 0.0001979667781446381,
296
+ "loss": 0.1312,
297
+ "step": 48
298
+ },
299
+ {
300
+ "epoch": 0.08,
301
+ "learning_rate": 0.00019785874696801202,
302
+ "loss": 0.1335,
303
+ "step": 49
304
+ },
305
+ {
306
+ "epoch": 0.08,
307
+ "learning_rate": 0.00019774795061135752,
308
+ "loss": 0.121,
309
+ "step": 50
310
+ },
311
+ {
312
+ "epoch": 0.08,
313
+ "learning_rate": 0.00019763439220543084,
314
+ "loss": 0.1268,
315
+ "step": 51
316
+ },
317
+ {
318
+ "epoch": 0.09,
319
+ "learning_rate": 0.00019751807495903484,
320
+ "loss": 0.1353,
321
+ "step": 52
322
+ },
323
+ {
324
+ "epoch": 0.09,
325
+ "learning_rate": 0.00019739900215892867,
326
+ "loss": 0.1242,
327
+ "step": 53
328
+ },
329
+ {
330
+ "epoch": 0.09,
331
+ "learning_rate": 0.0001972771771697347,
332
+ "loss": 0.1048,
333
+ "step": 54
334
+ },
335
+ {
336
+ "epoch": 0.09,
337
+ "learning_rate": 0.00019715260343384347,
338
+ "loss": 0.1329,
339
+ "step": 55
340
+ },
341
+ {
342
+ "epoch": 0.09,
343
+ "learning_rate": 0.00019702528447131646,
344
+ "loss": 0.1197,
345
+ "step": 56
346
+ },
347
+ {
348
+ "epoch": 0.09,
349
+ "learning_rate": 0.00019689522387978666,
350
+ "loss": 0.1463,
351
+ "step": 57
352
+ },
353
+ {
354
+ "epoch": 0.1,
355
+ "learning_rate": 0.00019676242533435678,
356
+ "loss": 0.1296,
357
+ "step": 58
358
+ },
359
+ {
360
+ "epoch": 0.1,
361
+ "learning_rate": 0.00019662689258749554,
362
+ "loss": 0.1243,
363
+ "step": 59
364
+ },
365
+ {
366
+ "epoch": 0.1,
367
+ "learning_rate": 0.00019648862946893158,
368
+ "loss": 0.1257,
369
+ "step": 60
370
+ },
371
+ {
372
+ "epoch": 0.1,
373
+ "learning_rate": 0.00019634763988554522,
374
+ "loss": 0.1348,
375
+ "step": 61
376
+ },
377
+ {
378
+ "epoch": 0.1,
379
+ "learning_rate": 0.0001962039278212581,
380
+ "loss": 0.1128,
381
+ "step": 62
382
+ },
383
+ {
384
+ "epoch": 0.1,
385
+ "learning_rate": 0.00019605749733692064,
386
+ "loss": 0.1227,
387
+ "step": 63
388
+ },
389
+ {
390
+ "epoch": 0.11,
391
+ "learning_rate": 0.00019590835257019714,
392
+ "loss": 0.1329,
393
+ "step": 64
394
+ },
395
+ {
396
+ "epoch": 0.11,
397
+ "learning_rate": 0.00019575649773544913,
398
+ "loss": 0.1291,
399
+ "step": 65
400
+ },
401
+ {
402
+ "epoch": 0.11,
403
+ "learning_rate": 0.00019560193712361596,
404
+ "loss": 0.119,
405
+ "step": 66
406
+ },
407
+ {
408
+ "epoch": 0.11,
409
+ "learning_rate": 0.00019544467510209388,
410
+ "loss": 0.126,
411
+ "step": 67
412
+ },
413
+ {
414
+ "epoch": 0.11,
415
+ "learning_rate": 0.00019528471611461235,
416
+ "loss": 0.1158,
417
+ "step": 68
418
+ },
419
+ {
420
+ "epoch": 0.11,
421
+ "learning_rate": 0.00019512206468110863,
422
+ "loss": 0.1309,
423
+ "step": 69
424
+ },
425
+ {
426
+ "epoch": 0.12,
427
+ "learning_rate": 0.00019495672539760007,
428
+ "loss": 0.1203,
429
+ "step": 70
430
+ },
431
+ {
432
+ "epoch": 0.12,
433
+ "learning_rate": 0.00019478870293605416,
434
+ "loss": 0.107,
435
+ "step": 71
436
+ },
437
+ {
438
+ "epoch": 0.12,
439
+ "learning_rate": 0.0001946180020442565,
440
+ "loss": 0.1249,
441
+ "step": 72
442
+ },
443
+ {
444
+ "epoch": 0.12,
445
+ "learning_rate": 0.00019444462754567682,
446
+ "loss": 0.1163,
447
+ "step": 73
448
+ },
449
+ {
450
+ "epoch": 0.12,
451
+ "learning_rate": 0.00019426858433933247,
452
+ "loss": 0.1192,
453
+ "step": 74
454
+ },
455
+ {
456
+ "epoch": 0.12,
457
+ "learning_rate": 0.00019408987739965005,
458
+ "loss": 0.1205,
459
+ "step": 75
460
+ },
461
+ {
462
+ "epoch": 0.13,
463
+ "learning_rate": 0.00019390851177632497,
464
+ "loss": 0.1205,
465
+ "step": 76
466
+ },
467
+ {
468
+ "epoch": 0.13,
469
+ "learning_rate": 0.00019372449259417857,
470
+ "loss": 0.1247,
471
+ "step": 77
472
+ },
473
+ {
474
+ "epoch": 0.13,
475
+ "learning_rate": 0.0001935378250530135,
476
+ "loss": 0.1195,
477
+ "step": 78
478
+ },
479
+ {
480
+ "epoch": 0.13,
481
+ "learning_rate": 0.00019334851442746664,
482
+ "loss": 0.1131,
483
+ "step": 79
484
+ },
485
+ {
486
+ "epoch": 0.13,
487
+ "learning_rate": 0.00019315656606686013,
488
+ "loss": 0.1119,
489
+ "step": 80
490
+ },
491
+ {
492
+ "epoch": 0.13,
493
+ "learning_rate": 0.00019296198539505013,
494
+ "loss": 0.1137,
495
+ "step": 81
496
+ },
497
+ {
498
+ "epoch": 0.14,
499
+ "learning_rate": 0.00019276477791027374,
500
+ "loss": 0.1052,
501
+ "step": 82
502
+ },
503
+ {
504
+ "epoch": 0.14,
505
+ "learning_rate": 0.00019256494918499346,
506
+ "loss": 0.1204,
507
+ "step": 83
508
+ },
509
+ {
510
+ "epoch": 0.14,
511
+ "learning_rate": 0.00019236250486573978,
512
+ "loss": 0.1158,
513
+ "step": 84
514
+ },
515
+ {
516
+ "epoch": 0.14,
517
+ "learning_rate": 0.00019215745067295169,
518
+ "loss": 0.1151,
519
+ "step": 85
520
+ },
521
+ {
522
+ "epoch": 0.14,
523
+ "learning_rate": 0.0001919497924008149,
524
+ "loss": 0.1274,
525
+ "step": 86
526
+ },
527
+ {
528
+ "epoch": 0.14,
529
+ "learning_rate": 0.00019173953591709828,
530
+ "loss": 0.1229,
531
+ "step": 87
532
+ },
533
+ {
534
+ "epoch": 0.15,
535
+ "learning_rate": 0.000191526687162988,
536
+ "loss": 0.1194,
537
+ "step": 88
538
+ },
539
+ {
540
+ "epoch": 0.15,
541
+ "learning_rate": 0.0001913112521529195,
542
+ "loss": 0.1409,
543
+ "step": 89
544
+ },
545
+ {
546
+ "epoch": 0.15,
547
+ "learning_rate": 0.00019109323697440782,
548
+ "loss": 0.1274,
549
+ "step": 90
550
+ },
551
+ {
552
+ "epoch": 0.15,
553
+ "learning_rate": 0.00019087264778787534,
554
+ "loss": 0.1188,
555
+ "step": 91
556
+ },
557
+ {
558
+ "epoch": 0.15,
559
+ "learning_rate": 0.00019064949082647786,
560
+ "loss": 0.1149,
561
+ "step": 92
562
+ },
563
+ {
564
+ "epoch": 0.15,
565
+ "learning_rate": 0.0001904237723959283,
566
+ "loss": 0.115,
567
+ "step": 93
568
+ },
569
+ {
570
+ "epoch": 0.16,
571
+ "learning_rate": 0.00019019549887431877,
572
+ "loss": 0.1297,
573
+ "step": 94
574
+ },
575
+ {
576
+ "epoch": 0.16,
577
+ "learning_rate": 0.00018996467671194016,
578
+ "loss": 0.1286,
579
+ "step": 95
580
+ },
581
+ {
582
+ "epoch": 0.16,
583
+ "learning_rate": 0.00018973131243109988,
584
+ "loss": 0.1061,
585
+ "step": 96
586
+ },
587
+ {
588
+ "epoch": 0.16,
589
+ "learning_rate": 0.00018949541262593762,
590
+ "loss": 0.1189,
591
+ "step": 97
592
+ },
593
+ {
594
+ "epoch": 0.16,
595
+ "learning_rate": 0.00018925698396223909,
596
+ "loss": 0.1372,
597
+ "step": 98
598
+ },
599
+ {
600
+ "epoch": 0.16,
601
+ "learning_rate": 0.0001890160331772474,
602
+ "loss": 0.118,
603
+ "step": 99
604
+ },
605
+ {
606
+ "epoch": 0.17,
607
+ "learning_rate": 0.00018877256707947306,
608
+ "loss": 0.1109,
609
+ "step": 100
610
+ },
611
+ {
612
+ "epoch": 0.17,
613
+ "learning_rate": 0.00018852659254850126,
614
+ "loss": 0.1142,
615
+ "step": 101
616
+ },
617
+ {
618
+ "epoch": 0.17,
619
+ "learning_rate": 0.00018827811653479768,
620
+ "loss": 0.1117,
621
+ "step": 102
622
+ },
623
+ {
624
+ "epoch": 0.17,
625
+ "learning_rate": 0.00018802714605951199,
626
+ "loss": 0.1251,
627
+ "step": 103
628
+ },
629
+ {
630
+ "epoch": 0.17,
631
+ "learning_rate": 0.00018777368821427953,
632
+ "loss": 0.1153,
633
+ "step": 104
634
+ },
635
+ {
636
+ "epoch": 0.17,
637
+ "learning_rate": 0.00018751775016102087,
638
+ "loss": 0.1288,
639
+ "step": 105
640
+ },
641
+ {
642
+ "epoch": 0.18,
643
+ "learning_rate": 0.00018725933913173938,
644
+ "loss": 0.1261,
645
+ "step": 106
646
+ },
647
+ {
648
+ "epoch": 0.18,
649
+ "learning_rate": 0.00018699846242831706,
650
+ "loss": 0.1349,
651
+ "step": 107
652
+ },
653
+ {
654
+ "epoch": 0.18,
655
+ "learning_rate": 0.00018673512742230802,
656
+ "loss": 0.1145,
657
+ "step": 108
658
+ },
659
+ {
660
+ "epoch": 0.18,
661
+ "learning_rate": 0.00018646934155473022,
662
+ "loss": 0.1206,
663
+ "step": 109
664
+ },
665
+ {
666
+ "epoch": 0.18,
667
+ "learning_rate": 0.0001862011123358554,
668
+ "loss": 0.1169,
669
+ "step": 110
670
+ },
671
+ {
672
+ "epoch": 0.18,
673
+ "learning_rate": 0.00018593044734499655,
674
+ "loss": 0.1256,
675
+ "step": 111
676
+ },
677
+ {
678
+ "epoch": 0.19,
679
+ "learning_rate": 0.00018565735423029404,
680
+ "loss": 0.1233,
681
+ "step": 112
682
+ },
683
+ {
684
+ "epoch": 0.19,
685
+ "learning_rate": 0.00018538184070849924,
686
+ "loss": 0.1252,
687
+ "step": 113
688
+ },
689
+ {
690
+ "epoch": 0.19,
691
+ "learning_rate": 0.00018510391456475676,
692
+ "loss": 0.1174,
693
+ "step": 114
694
+ },
695
+ {
696
+ "epoch": 0.19,
697
+ "learning_rate": 0.00018482358365238413,
698
+ "loss": 0.1205,
699
+ "step": 115
700
+ },
701
+ {
702
+ "epoch": 0.19,
703
+ "learning_rate": 0.0001845408558926502,
704
+ "loss": 0.1298,
705
+ "step": 116
706
+ },
707
+ {
708
+ "epoch": 0.19,
709
+ "learning_rate": 0.00018425573927455117,
710
+ "loss": 0.1085,
711
+ "step": 117
712
+ },
713
+ {
714
+ "epoch": 0.2,
715
+ "learning_rate": 0.0001839682418545848,
716
+ "loss": 0.1363,
717
+ "step": 118
718
+ },
719
+ {
720
+ "epoch": 0.2,
721
+ "learning_rate": 0.00018367837175652284,
722
+ "loss": 0.1228,
723
+ "step": 119
724
+ },
725
+ {
726
+ "epoch": 0.2,
727
+ "learning_rate": 0.0001833861371711814,
728
+ "loss": 0.1259,
729
+ "step": 120
730
+ },
731
+ {
732
+ "epoch": 0.2,
733
+ "learning_rate": 0.00018309154635618965,
734
+ "loss": 0.1246,
735
+ "step": 121
736
+ },
737
+ {
738
+ "epoch": 0.2,
739
+ "learning_rate": 0.00018279460763575637,
740
+ "loss": 0.1288,
741
+ "step": 122
742
+ },
743
+ {
744
+ "epoch": 0.2,
745
+ "learning_rate": 0.0001824953294004347,
746
+ "loss": 0.1203,
747
+ "step": 123
748
+ },
749
+ {
750
+ "epoch": 0.21,
751
+ "learning_rate": 0.00018219372010688515,
752
+ "loss": 0.1166,
753
+ "step": 124
754
+ },
755
+ {
756
+ "epoch": 0.21,
757
+ "learning_rate": 0.00018188978827763652,
758
+ "loss": 0.1377,
759
+ "step": 125
760
+ },
761
+ {
762
+ "epoch": 0.21,
763
+ "learning_rate": 0.00018158354250084527,
764
+ "loss": 0.1364,
765
+ "step": 126
766
+ },
767
+ {
768
+ "epoch": 0.21,
769
+ "learning_rate": 0.00018127499143005268,
770
+ "loss": 0.1179,
771
+ "step": 127
772
+ },
773
+ {
774
+ "epoch": 0.21,
775
+ "learning_rate": 0.00018096414378394028,
776
+ "loss": 0.1308,
777
+ "step": 128
778
+ },
779
+ {
780
+ "epoch": 0.21,
781
+ "learning_rate": 0.00018065100834608377,
782
+ "loss": 0.1359,
783
+ "step": 129
784
+ },
785
+ {
786
+ "epoch": 0.22,
787
+ "learning_rate": 0.00018033559396470454,
788
+ "loss": 0.1253,
789
+ "step": 130
790
+ },
791
+ {
792
+ "epoch": 0.22,
793
+ "learning_rate": 0.00018001790955241972,
794
+ "loss": 0.1106,
795
+ "step": 131
796
+ },
797
+ {
798
+ "epoch": 0.22,
799
+ "learning_rate": 0.0001796979640859904,
800
+ "loss": 0.118,
801
+ "step": 132
802
+ },
803
+ {
804
+ "epoch": 0.22,
805
+ "learning_rate": 0.000179375766606068,
806
+ "loss": 0.1361,
807
+ "step": 133
808
+ },
809
+ {
810
+ "epoch": 0.22,
811
+ "learning_rate": 0.0001790513262169386,
812
+ "loss": 0.1388,
813
+ "step": 134
814
+ },
815
+ {
816
+ "epoch": 0.22,
817
+ "learning_rate": 0.00017872465208626598,
818
+ "loss": 0.1448,
819
+ "step": 135
820
+ },
821
+ {
822
+ "epoch": 0.23,
823
+ "learning_rate": 0.00017839575344483238,
824
+ "loss": 0.111,
825
+ "step": 136
826
+ },
827
+ {
828
+ "epoch": 0.23,
829
+ "learning_rate": 0.00017806463958627762,
830
+ "loss": 0.115,
831
+ "step": 137
832
+ },
833
+ {
834
+ "epoch": 0.23,
835
+ "learning_rate": 0.00017773131986683672,
836
+ "loss": 0.1198,
837
+ "step": 138
838
+ },
839
+ {
840
+ "epoch": 0.23,
841
+ "learning_rate": 0.00017739580370507532,
842
+ "loss": 0.1347,
843
+ "step": 139
844
+ },
845
+ {
846
+ "epoch": 0.23,
847
+ "learning_rate": 0.00017705810058162353,
848
+ "loss": 0.12,
849
+ "step": 140
850
+ },
851
+ {
852
+ "epoch": 0.23,
853
+ "learning_rate": 0.00017671822003890823,
854
+ "loss": 0.1258,
855
+ "step": 141
856
+ },
857
+ {
858
+ "epoch": 0.24,
859
+ "learning_rate": 0.00017637617168088325,
860
+ "loss": 0.1357,
861
+ "step": 142
862
+ },
863
+ {
864
+ "epoch": 0.24,
865
+ "learning_rate": 0.0001760319651727581,
866
+ "loss": 0.132,
867
+ "step": 143
868
+ },
869
+ {
870
+ "epoch": 0.24,
871
+ "learning_rate": 0.0001756856102407247,
872
+ "loss": 0.1134,
873
+ "step": 144
874
+ },
875
+ {
876
+ "epoch": 0.24,
877
+ "learning_rate": 0.0001753371166716828,
878
+ "loss": 0.1106,
879
+ "step": 145
880
+ },
881
+ {
882
+ "epoch": 0.24,
883
+ "learning_rate": 0.00017498649431296322,
884
+ "loss": 0.1133,
885
+ "step": 146
886
+ },
887
+ {
888
+ "epoch": 0.24,
889
+ "learning_rate": 0.0001746337530720497,
890
+ "loss": 0.1232,
891
+ "step": 147
892
+ },
893
+ {
894
+ "epoch": 0.25,
895
+ "learning_rate": 0.00017427890291629893,
896
+ "loss": 0.1249,
897
+ "step": 148
898
+ },
899
+ {
900
+ "epoch": 0.25,
901
+ "learning_rate": 0.00017392195387265887,
902
+ "loss": 0.1172,
903
+ "step": 149
904
+ },
905
+ {
906
+ "epoch": 0.25,
907
+ "learning_rate": 0.00017356291602738542,
908
+ "loss": 0.1112,
909
+ "step": 150
910
+ },
911
+ {
912
+ "epoch": 0.25,
913
+ "learning_rate": 0.0001732017995257575,
914
+ "loss": 0.1124,
915
+ "step": 151
916
+ },
917
+ {
918
+ "epoch": 0.25,
919
+ "learning_rate": 0.00017283861457179022,
920
+ "loss": 0.124,
921
+ "step": 152
922
+ },
923
+ {
924
+ "epoch": 0.25,
925
+ "learning_rate": 0.00017247337142794678,
926
+ "loss": 0.1116,
927
+ "step": 153
928
+ },
929
+ {
930
+ "epoch": 0.26,
931
+ "learning_rate": 0.0001721060804148482,
932
+ "loss": 0.1294,
933
+ "step": 154
934
+ },
935
+ {
936
+ "epoch": 0.26,
937
+ "learning_rate": 0.0001717367519109819,
938
+ "loss": 0.1185,
939
+ "step": 155
940
+ },
941
+ {
942
+ "epoch": 0.26,
943
+ "learning_rate": 0.00017136539635240837,
944
+ "loss": 0.1383,
945
+ "step": 156
946
+ },
947
+ {
948
+ "epoch": 0.26,
949
+ "learning_rate": 0.0001709920242324663,
950
+ "loss": 0.1204,
951
+ "step": 157
952
+ },
953
+ {
954
+ "epoch": 0.26,
955
+ "learning_rate": 0.00017061664610147604,
956
+ "loss": 0.1293,
957
+ "step": 158
958
+ },
959
+ {
960
+ "epoch": 0.26,
961
+ "learning_rate": 0.0001702392725664415,
962
+ "loss": 0.1303,
963
+ "step": 159
964
+ },
965
+ {
966
+ "epoch": 0.27,
967
+ "learning_rate": 0.00016985991429075036,
968
+ "loss": 0.1189,
969
+ "step": 160
970
+ },
971
+ {
972
+ "epoch": 0.27,
973
+ "learning_rate": 0.00016947858199387294,
974
+ "loss": 0.1431,
975
+ "step": 161
976
+ },
977
+ {
978
+ "epoch": 0.27,
979
+ "learning_rate": 0.00016909528645105907,
980
+ "loss": 0.1174,
981
+ "step": 162
982
+ },
983
+ {
984
+ "epoch": 0.27,
985
+ "learning_rate": 0.00016871003849303382,
986
+ "loss": 0.1306,
987
+ "step": 163
988
+ },
989
+ {
990
+ "epoch": 0.27,
991
+ "learning_rate": 0.0001683228490056913,
992
+ "loss": 0.1086,
993
+ "step": 164
994
+ },
995
+ {
996
+ "epoch": 0.27,
997
+ "learning_rate": 0.00016793372892978713,
998
+ "loss": 0.122,
999
+ "step": 165
1000
+ },
1001
+ {
1002
+ "epoch": 0.28,
1003
+ "learning_rate": 0.00016754268926062938,
1004
+ "loss": 0.1081,
1005
+ "step": 166
1006
+ },
1007
+ {
1008
+ "epoch": 0.28,
1009
+ "learning_rate": 0.0001671497410477676,
1010
+ "loss": 0.1184,
1011
+ "step": 167
1012
+ },
1013
+ {
1014
+ "epoch": 0.28,
1015
+ "learning_rate": 0.00016675489539468092,
1016
+ "loss": 0.1238,
1017
+ "step": 168
1018
+ },
1019
+ {
1020
+ "epoch": 0.28,
1021
+ "learning_rate": 0.0001663581634584641,
1022
+ "loss": 0.1256,
1023
+ "step": 169
1024
+ },
1025
+ {
1026
+ "epoch": 0.28,
1027
+ "learning_rate": 0.0001659595564495124,
1028
+ "loss": 0.1242,
1029
+ "step": 170
1030
+ },
1031
+ {
1032
+ "epoch": 0.28,
1033
+ "learning_rate": 0.00016555908563120457,
1034
+ "loss": 0.1179,
1035
+ "step": 171
1036
+ },
1037
+ {
1038
+ "epoch": 0.29,
1039
+ "learning_rate": 0.0001651567623195849,
1040
+ "loss": 0.1173,
1041
+ "step": 172
1042
+ },
1043
+ {
1044
+ "epoch": 0.29,
1045
+ "learning_rate": 0.00016475259788304317,
1046
+ "loss": 0.1146,
1047
+ "step": 173
1048
+ },
1049
+ {
1050
+ "epoch": 0.29,
1051
+ "learning_rate": 0.00016434660374199376,
1052
+ "loss": 0.1254,
1053
+ "step": 174
1054
+ },
1055
+ {
1056
+ "epoch": 0.29,
1057
+ "learning_rate": 0.00016393879136855248,
1058
+ "loss": 0.1263,
1059
+ "step": 175
1060
+ },
1061
+ {
1062
+ "epoch": 0.29,
1063
+ "learning_rate": 0.00016352917228621284,
1064
+ "loss": 0.1181,
1065
+ "step": 176
1066
+ },
1067
+ {
1068
+ "epoch": 0.29,
1069
+ "learning_rate": 0.0001631177580695202,
1070
+ "loss": 0.1213,
1071
+ "step": 177
1072
+ },
1073
+ {
1074
+ "epoch": 0.3,
1075
+ "learning_rate": 0.00016270456034374474,
1076
+ "loss": 0.1156,
1077
+ "step": 178
1078
+ },
1079
+ {
1080
+ "epoch": 0.3,
1081
+ "learning_rate": 0.00016228959078455306,
1082
+ "loss": 0.1145,
1083
+ "step": 179
1084
+ },
1085
+ {
1086
+ "epoch": 0.3,
1087
+ "learning_rate": 0.0001618728611176781,
1088
+ "loss": 0.1385,
1089
+ "step": 180
1090
+ },
1091
+ {
1092
+ "epoch": 0.3,
1093
+ "learning_rate": 0.000161454383118588,
1094
+ "loss": 0.1228,
1095
+ "step": 181
1096
+ },
1097
+ {
1098
+ "epoch": 0.3,
1099
+ "learning_rate": 0.00016103416861215313,
1100
+ "loss": 0.1175,
1101
+ "step": 182
1102
+ },
1103
+ {
1104
+ "epoch": 0.3,
1105
+ "learning_rate": 0.00016061222947231225,
1106
+ "loss": 0.1328,
1107
+ "step": 183
1108
+ },
1109
+ {
1110
+ "epoch": 0.31,
1111
+ "learning_rate": 0.0001601885776217367,
1112
+ "loss": 0.1223,
1113
+ "step": 184
1114
+ },
1115
+ {
1116
+ "epoch": 0.31,
1117
+ "learning_rate": 0.00015976322503149373,
1118
+ "loss": 0.1244,
1119
+ "step": 185
1120
+ },
1121
+ {
1122
+ "epoch": 0.31,
1123
+ "learning_rate": 0.00015933618372070805,
1124
+ "loss": 0.1356,
1125
+ "step": 186
1126
+ },
1127
+ {
1128
+ "epoch": 0.31,
1129
+ "learning_rate": 0.00015890746575622231,
1130
+ "loss": 0.12,
1131
+ "step": 187
1132
+ },
1133
+ {
1134
+ "epoch": 0.31,
1135
+ "learning_rate": 0.00015847708325225618,
1136
+ "loss": 0.1254,
1137
+ "step": 188
1138
+ },
1139
+ {
1140
+ "epoch": 0.31,
1141
+ "learning_rate": 0.00015804504837006394,
1142
+ "loss": 0.1139,
1143
+ "step": 189
1144
+ },
1145
+ {
1146
+ "epoch": 0.32,
1147
+ "learning_rate": 0.00015761137331759084,
1148
+ "loss": 0.1251,
1149
+ "step": 190
1150
+ },
1151
+ {
1152
+ "epoch": 0.32,
1153
+ "learning_rate": 0.0001571760703491282,
1154
+ "loss": 0.1269,
1155
+ "step": 191
1156
+ },
1157
+ {
1158
+ "epoch": 0.32,
1159
+ "learning_rate": 0.00015673915176496713,
1160
+ "loss": 0.1203,
1161
+ "step": 192
1162
+ },
1163
+ {
1164
+ "epoch": 0.32,
1165
+ "learning_rate": 0.00015630062991105098,
1166
+ "loss": 0.108,
1167
+ "step": 193
1168
+ },
1169
+ {
1170
+ "epoch": 0.32,
1171
+ "learning_rate": 0.00015586051717862636,
1172
+ "loss": 0.1276,
1173
+ "step": 194
1174
+ },
1175
+ {
1176
+ "epoch": 0.32,
1177
+ "learning_rate": 0.0001554188260038932,
1178
+ "loss": 0.1164,
1179
+ "step": 195
1180
+ },
1181
+ {
1182
+ "epoch": 0.33,
1183
+ "learning_rate": 0.00015497556886765316,
1184
+ "loss": 0.1212,
1185
+ "step": 196
1186
+ },
1187
+ {
1188
+ "epoch": 0.33,
1189
+ "learning_rate": 0.0001545307582949571,
1190
+ "loss": 0.1358,
1191
+ "step": 197
1192
+ },
1193
+ {
1194
+ "epoch": 0.33,
1195
+ "learning_rate": 0.00015408440685475109,
1196
+ "loss": 0.1168,
1197
+ "step": 198
1198
+ },
1199
+ {
1200
+ "epoch": 0.33,
1201
+ "learning_rate": 0.0001536365271595212,
1202
+ "loss": 0.1237,
1203
+ "step": 199
1204
+ },
1205
+ {
1206
+ "epoch": 0.33,
1207
+ "learning_rate": 0.00015318713186493734,
1208
+ "loss": 0.136,
1209
+ "step": 200
1210
+ },
1211
+ {
1212
+ "epoch": 0.33,
1213
+ "learning_rate": 0.00015273623366949523,
1214
+ "loss": 0.1332,
1215
+ "step": 201
1216
+ },
1217
+ {
1218
+ "epoch": 0.34,
1219
+ "learning_rate": 0.0001522838453141581,
1220
+ "loss": 0.1122,
1221
+ "step": 202
1222
+ },
1223
+ {
1224
+ "epoch": 0.34,
1225
+ "learning_rate": 0.00015182997958199617,
1226
+ "loss": 0.118,
1227
+ "step": 203
1228
+ },
1229
+ {
1230
+ "epoch": 0.34,
1231
+ "learning_rate": 0.00015137464929782586,
1232
+ "loss": 0.1277,
1233
+ "step": 204
1234
+ },
1235
+ {
1236
+ "epoch": 0.34,
1237
+ "learning_rate": 0.00015091786732784716,
1238
+ "loss": 0.1117,
1239
+ "step": 205
1240
+ },
1241
+ {
1242
+ "epoch": 0.34,
1243
+ "learning_rate": 0.00015045964657928006,
1244
+ "loss": 0.1025,
1245
+ "step": 206
1246
+ },
1247
+ {
1248
+ "epoch": 0.34,
1249
+ "learning_rate": 0.00015000000000000001,
1250
+ "loss": 0.1251,
1251
+ "step": 207
1252
+ },
1253
+ {
1254
+ "epoch": 0.35,
1255
+ "learning_rate": 0.00014953894057817188,
1256
+ "loss": 0.142,
1257
+ "step": 208
1258
+ },
1259
+ {
1260
+ "epoch": 0.35,
1261
+ "learning_rate": 0.00014907648134188304,
1262
+ "loss": 0.1322,
1263
+ "step": 209
1264
+ },
1265
+ {
1266
+ "epoch": 0.35,
1267
+ "learning_rate": 0.0001486126353587752,
1268
+ "loss": 0.1236,
1269
+ "step": 210
1270
+ },
1271
+ {
1272
+ "epoch": 0.35,
1273
+ "learning_rate": 0.00014814741573567514,
1274
+ "loss": 0.1293,
1275
+ "step": 211
1276
+ },
1277
+ {
1278
+ "epoch": 0.35,
1279
+ "learning_rate": 0.0001476808356182245,
1280
+ "loss": 0.1157,
1281
+ "step": 212
1282
+ },
1283
+ {
1284
+ "epoch": 0.35,
1285
+ "learning_rate": 0.00014721290819050804,
1286
+ "loss": 0.1299,
1287
+ "step": 213
1288
+ },
1289
+ {
1290
+ "epoch": 0.36,
1291
+ "learning_rate": 0.0001467436466746814,
1292
+ "loss": 0.1052,
1293
+ "step": 214
1294
+ },
1295
+ {
1296
+ "epoch": 0.36,
1297
+ "learning_rate": 0.00014627306433059723,
1298
+ "loss": 0.1242,
1299
+ "step": 215
1300
+ },
1301
+ {
1302
+ "epoch": 0.36,
1303
+ "learning_rate": 0.00014580117445543077,
1304
+ "loss": 0.103,
1305
+ "step": 216
1306
+ },
1307
+ {
1308
+ "epoch": 0.36,
1309
+ "learning_rate": 0.00014532799038330385,
1310
+ "loss": 0.1274,
1311
+ "step": 217
1312
+ },
1313
+ {
1314
+ "epoch": 0.36,
1315
+ "learning_rate": 0.00014485352548490826,
1316
+ "loss": 0.12,
1317
+ "step": 218
1318
+ },
1319
+ {
1320
+ "epoch": 0.36,
1321
+ "learning_rate": 0.00014437779316712796,
1322
+ "loss": 0.1113,
1323
+ "step": 219
1324
+ },
1325
+ {
1326
+ "epoch": 0.37,
1327
+ "learning_rate": 0.00014390080687266013,
1328
+ "loss": 0.1244,
1329
+ "step": 220
1330
+ },
1331
+ {
1332
+ "epoch": 0.37,
1333
+ "learning_rate": 0.0001434225800796354,
1334
+ "loss": 0.1111,
1335
+ "step": 221
1336
+ },
1337
+ {
1338
+ "epoch": 0.37,
1339
+ "learning_rate": 0.000142943126301237,
1340
+ "loss": 0.1007,
1341
+ "step": 222
1342
+ },
1343
+ {
1344
+ "epoch": 0.37,
1345
+ "learning_rate": 0.00014246245908531882,
1346
+ "loss": 0.1175,
1347
+ "step": 223
1348
+ },
1349
+ {
1350
+ "epoch": 0.37,
1351
+ "learning_rate": 0.00014198059201402287,
1352
+ "loss": 0.1285,
1353
+ "step": 224
1354
+ },
1355
+ {
1356
+ "epoch": 0.37,
1357
+ "learning_rate": 0.00014149753870339507,
1358
+ "loss": 0.1523,
1359
+ "step": 225
1360
+ },
1361
+ {
1362
+ "epoch": 0.38,
1363
+ "learning_rate": 0.0001410133128030009,
1364
+ "loss": 0.1212,
1365
+ "step": 226
1366
+ },
1367
+ {
1368
+ "epoch": 0.38,
1369
+ "learning_rate": 0.00014052792799553934,
1370
+ "loss": 0.1053,
1371
+ "step": 227
1372
+ },
1373
+ {
1374
+ "epoch": 0.38,
1375
+ "learning_rate": 0.00014004139799645668,
1376
+ "loss": 0.1356,
1377
+ "step": 228
1378
+ },
1379
+ {
1380
+ "epoch": 0.38,
1381
+ "learning_rate": 0.0001395537365535585,
1382
+ "loss": 0.1316,
1383
+ "step": 229
1384
+ },
1385
+ {
1386
+ "epoch": 0.38,
1387
+ "learning_rate": 0.00013906495744662157,
1388
+ "loss": 0.1316,
1389
+ "step": 230
1390
+ },
1391
+ {
1392
+ "epoch": 0.38,
1393
+ "learning_rate": 0.00013857507448700423,
1394
+ "loss": 0.1209,
1395
+ "step": 231
1396
+ },
1397
+ {
1398
+ "epoch": 0.39,
1399
+ "learning_rate": 0.0001380841015172563,
1400
+ "loss": 0.1191,
1401
+ "step": 232
1402
+ },
1403
+ {
1404
+ "epoch": 0.39,
1405
+ "learning_rate": 0.00013759205241072782,
1406
+ "loss": 0.1207,
1407
+ "step": 233
1408
+ },
1409
+ {
1410
+ "epoch": 0.39,
1411
+ "learning_rate": 0.00013709894107117698,
1412
+ "loss": 0.1234,
1413
+ "step": 234
1414
+ },
1415
+ {
1416
+ "epoch": 0.39,
1417
+ "learning_rate": 0.00013660478143237746,
1418
+ "loss": 0.1269,
1419
+ "step": 235
1420
+ },
1421
+ {
1422
+ "epoch": 0.39,
1423
+ "learning_rate": 0.00013610958745772456,
1424
+ "loss": 0.1163,
1425
+ "step": 236
1426
+ },
1427
+ {
1428
+ "epoch": 0.39,
1429
+ "learning_rate": 0.00013561337313984054,
1430
+ "loss": 0.1369,
1431
+ "step": 237
1432
+ },
1433
+ {
1434
+ "epoch": 0.4,
1435
+ "learning_rate": 0.0001351161525001795,
1436
+ "loss": 0.1146,
1437
+ "step": 238
1438
+ },
1439
+ {
1440
+ "epoch": 0.4,
1441
+ "learning_rate": 0.00013461793958863087,
1442
+ "loss": 0.1078,
1443
+ "step": 239
1444
+ },
1445
+ {
1446
+ "epoch": 0.4,
1447
+ "learning_rate": 0.00013411874848312272,
1448
+ "loss": 0.1238,
1449
+ "step": 240
1450
+ },
1451
+ {
1452
+ "epoch": 0.4,
1453
+ "learning_rate": 0.0001336185932892237,
1454
+ "loss": 0.1116,
1455
+ "step": 241
1456
+ },
1457
+ {
1458
+ "epoch": 0.4,
1459
+ "learning_rate": 0.00013311748813974453,
1460
+ "loss": 0.1365,
1461
+ "step": 242
1462
+ },
1463
+ {
1464
+ "epoch": 0.4,
1465
+ "learning_rate": 0.0001326154471943388,
1466
+ "loss": 0.1286,
1467
+ "step": 243
1468
+ },
1469
+ {
1470
+ "epoch": 0.41,
1471
+ "learning_rate": 0.00013211248463910262,
1472
+ "loss": 0.1142,
1473
+ "step": 244
1474
+ },
1475
+ {
1476
+ "epoch": 0.41,
1477
+ "learning_rate": 0.000131608614686174,
1478
+ "loss": 0.1158,
1479
+ "step": 245
1480
+ },
1481
+ {
1482
+ "epoch": 0.41,
1483
+ "learning_rate": 0.0001311038515733311,
1484
+ "loss": 1.5948,
1485
+ "step": 246
1486
+ },
1487
+ {
1488
+ "epoch": 0.41,
1489
+ "learning_rate": 0.00013059820956358998,
1490
+ "loss": 0.1267,
1491
+ "step": 247
1492
+ },
1493
+ {
1494
+ "epoch": 0.41,
1495
+ "learning_rate": 0.00013009170294480147,
1496
+ "loss": 0.1171,
1497
+ "step": 248
1498
+ },
1499
+ {
1500
+ "epoch": 0.41,
1501
+ "learning_rate": 0.0001295843460292477,
1502
+ "loss": 0.128,
1503
+ "step": 249
1504
+ },
1505
+ {
1506
+ "epoch": 0.42,
1507
+ "learning_rate": 0.0001290761531532374,
1508
+ "loss": 0.1455,
1509
+ "step": 250
1510
+ },
1511
+ {
1512
+ "epoch": 0.42,
1513
+ "learning_rate": 0.0001285671386767009,
1514
+ "loss": 0.1207,
1515
+ "step": 251
1516
+ },
1517
+ {
1518
+ "epoch": 0.42,
1519
+ "learning_rate": 0.00012805731698278442,
1520
+ "loss": 0.1266,
1521
+ "step": 252
1522
+ },
1523
+ {
1524
+ "epoch": 0.42,
1525
+ "learning_rate": 0.00012754670247744354,
1526
+ "loss": 0.125,
1527
+ "step": 253
1528
+ },
1529
+ {
1530
+ "epoch": 0.42,
1531
+ "learning_rate": 0.0001270353095890363,
1532
+ "loss": 0.1238,
1533
+ "step": 254
1534
+ },
1535
+ {
1536
+ "epoch": 0.42,
1537
+ "learning_rate": 0.00012652315276791528,
1538
+ "loss": 0.125,
1539
+ "step": 255
1540
+ },
1541
+ {
1542
+ "epoch": 0.43,
1543
+ "learning_rate": 0.0001260102464860195,
1544
+ "loss": 0.1201,
1545
+ "step": 256
1546
+ },
1547
+ {
1548
+ "epoch": 0.43,
1549
+ "learning_rate": 0.00012549660523646528,
1550
+ "loss": 0.1227,
1551
+ "step": 257
1552
+ },
1553
+ {
1554
+ "epoch": 0.43,
1555
+ "learning_rate": 0.00012498224353313684,
1556
+ "loss": 0.1205,
1557
+ "step": 258
1558
+ },
1559
+ {
1560
+ "epoch": 0.43,
1561
+ "learning_rate": 0.00012446717591027624,
1562
+ "loss": 0.1327,
1563
+ "step": 259
1564
+ },
1565
+ {
1566
+ "epoch": 0.43,
1567
+ "learning_rate": 0.00012395141692207243,
1568
+ "loss": 0.1193,
1569
+ "step": 260
1570
+ },
1571
+ {
1572
+ "epoch": 0.43,
1573
+ "learning_rate": 0.00012343498114225038,
1574
+ "loss": 0.114,
1575
+ "step": 261
1576
+ },
1577
+ {
1578
+ "epoch": 0.44,
1579
+ "learning_rate": 0.00012291788316365888,
1580
+ "loss": 0.1143,
1581
+ "step": 262
1582
+ },
1583
+ {
1584
+ "epoch": 0.44,
1585
+ "learning_rate": 0.00012240013759785848,
1586
+ "loss": 0.1168,
1587
+ "step": 263
1588
+ },
1589
+ {
1590
+ "epoch": 0.44,
1591
+ "learning_rate": 0.00012188175907470847,
1592
+ "loss": 0.1348,
1593
+ "step": 264
1594
+ },
1595
+ {
1596
+ "epoch": 0.44,
1597
+ "learning_rate": 0.00012136276224195348,
1598
+ "loss": 0.1153,
1599
+ "step": 265
1600
+ },
1601
+ {
1602
+ "epoch": 0.44,
1603
+ "learning_rate": 0.00012084316176480973,
1604
+ "loss": 0.1334,
1605
+ "step": 266
1606
+ },
1607
+ {
1608
+ "epoch": 0.44,
1609
+ "learning_rate": 0.00012032297232555039,
1610
+ "loss": 0.1024,
1611
+ "step": 267
1612
+ },
1613
+ {
1614
+ "epoch": 0.45,
1615
+ "learning_rate": 0.00011980220862309097,
1616
+ "loss": 0.1257,
1617
+ "step": 268
1618
+ },
1619
+ {
1620
+ "epoch": 0.45,
1621
+ "learning_rate": 0.00011928088537257375,
1622
+ "loss": 0.1145,
1623
+ "step": 269
1624
+ },
1625
+ {
1626
+ "epoch": 0.45,
1627
+ "learning_rate": 0.00011875901730495215,
1628
+ "loss": 0.1288,
1629
+ "step": 270
1630
+ },
1631
+ {
1632
+ "epoch": 0.45,
1633
+ "learning_rate": 0.0001182366191665744,
1634
+ "loss": 0.127,
1635
+ "step": 271
1636
+ },
1637
+ {
1638
+ "epoch": 0.45,
1639
+ "learning_rate": 0.00011771370571876681,
1640
+ "loss": 0.118,
1641
+ "step": 272
1642
+ },
1643
+ {
1644
+ "epoch": 0.45,
1645
+ "learning_rate": 0.00011719029173741676,
1646
+ "loss": 0.1216,
1647
+ "step": 273
1648
+ },
1649
+ {
1650
+ "epoch": 0.46,
1651
+ "learning_rate": 0.00011666639201255506,
1652
+ "loss": 0.1261,
1653
+ "step": 274
1654
+ },
1655
+ {
1656
+ "epoch": 0.46,
1657
+ "learning_rate": 0.00011614202134793823,
1658
+ "loss": 0.1207,
1659
+ "step": 275
1660
+ },
1661
+ {
1662
+ "epoch": 0.46,
1663
+ "learning_rate": 0.00011561719456062994,
1664
+ "loss": 0.1165,
1665
+ "step": 276
1666
+ },
1667
+ {
1668
+ "epoch": 0.46,
1669
+ "learning_rate": 0.00011509192648058249,
1670
+ "loss": 0.1293,
1671
+ "step": 277
1672
+ },
1673
+ {
1674
+ "epoch": 0.46,
1675
+ "learning_rate": 0.00011456623195021778,
1676
+ "loss": 0.1447,
1677
+ "step": 278
1678
+ },
1679
+ {
1680
+ "epoch": 0.46,
1681
+ "learning_rate": 0.00011404012582400779,
1682
+ "loss": 2.074,
1683
+ "step": 279
1684
+ },
1685
+ {
1686
+ "epoch": 0.47,
1687
+ "learning_rate": 0.00011351362296805485,
1688
+ "loss": 0.1065,
1689
+ "step": 280
1690
+ },
1691
+ {
1692
+ "epoch": 0.47,
1693
+ "learning_rate": 0.00011298673825967183,
1694
+ "loss": 0.1089,
1695
+ "step": 281
1696
+ },
1697
+ {
1698
+ "epoch": 0.47,
1699
+ "learning_rate": 0.00011245948658696126,
1700
+ "loss": 0.1108,
1701
+ "step": 282
1702
+ },
1703
+ {
1704
+ "epoch": 0.47,
1705
+ "learning_rate": 0.00011193188284839517,
1706
+ "loss": 0.1295,
1707
+ "step": 283
1708
+ },
1709
+ {
1710
+ "epoch": 0.47,
1711
+ "learning_rate": 0.00011140394195239376,
1712
+ "loss": 0.1173,
1713
+ "step": 284
1714
+ },
1715
+ {
1716
+ "epoch": 0.47,
1717
+ "learning_rate": 0.00011087567881690422,
1718
+ "loss": 0.1269,
1719
+ "step": 285
1720
+ },
1721
+ {
1722
+ "epoch": 0.48,
1723
+ "learning_rate": 0.00011034710836897921,
1724
+ "loss": 0.1198,
1725
+ "step": 286
1726
+ },
1727
+ {
1728
+ "epoch": 0.48,
1729
+ "learning_rate": 0.00010981824554435518,
1730
+ "loss": 0.153,
1731
+ "step": 287
1732
+ },
1733
+ {
1734
+ "epoch": 0.48,
1735
+ "learning_rate": 0.00010928910528703007,
1736
+ "loss": 0.1144,
1737
+ "step": 288
1738
+ },
1739
+ {
1740
+ "epoch": 0.48,
1741
+ "learning_rate": 0.0001087597025488413,
1742
+ "loss": 0.1291,
1743
+ "step": 289
1744
+ },
1745
+ {
1746
+ "epoch": 0.48,
1747
+ "learning_rate": 0.00010823005228904314,
1748
+ "loss": 0.1168,
1749
+ "step": 290
1750
+ },
1751
+ {
1752
+ "epoch": 0.48,
1753
+ "learning_rate": 0.00010770016947388407,
1754
+ "loss": 0.1037,
1755
+ "step": 291
1756
+ },
1757
+ {
1758
+ "epoch": 0.49,
1759
+ "learning_rate": 0.00010717006907618377,
1760
+ "loss": 0.1261,
1761
+ "step": 292
1762
+ },
1763
+ {
1764
+ "epoch": 0.49,
1765
+ "learning_rate": 0.0001066397660749102,
1766
+ "loss": 0.1297,
1767
+ "step": 293
1768
+ },
1769
+ {
1770
+ "epoch": 0.49,
1771
+ "learning_rate": 0.00010610927545475624,
1772
+ "loss": 0.1124,
1773
+ "step": 294
1774
+ },
1775
+ {
1776
+ "epoch": 0.49,
1777
+ "learning_rate": 0.00010557861220571625,
1778
+ "loss": 0.1249,
1779
+ "step": 295
1780
+ },
1781
+ {
1782
+ "epoch": 0.49,
1783
+ "learning_rate": 0.0001050477913226626,
1784
+ "loss": 0.1205,
1785
+ "step": 296
1786
+ },
1787
+ {
1788
+ "epoch": 0.49,
1789
+ "learning_rate": 0.00010451682780492189,
1790
+ "loss": 0.1174,
1791
+ "step": 297
1792
+ },
1793
+ {
1794
+ "epoch": 0.5,
1795
+ "learning_rate": 0.00010398573665585105,
1796
+ "loss": 0.1213,
1797
+ "step": 298
1798
+ },
1799
+ {
1800
+ "epoch": 0.5,
1801
+ "learning_rate": 0.00010345453288241356,
1802
+ "loss": 0.1079,
1803
+ "step": 299
1804
+ },
1805
+ {
1806
+ "epoch": 0.5,
1807
+ "learning_rate": 0.00010292323149475527,
1808
+ "loss": 0.1022,
1809
+ "step": 300
1810
+ },
1811
+ {
1812
+ "epoch": 0.5,
1813
+ "learning_rate": 0.0001023918475057803,
1814
+ "loss": 0.1013,
1815
+ "step": 301
1816
+ },
1817
+ {
1818
+ "epoch": 0.5,
1819
+ "learning_rate": 0.00010186039593072685,
1820
+ "loss": 0.1288,
1821
+ "step": 302
1822
+ },
1823
+ {
1824
+ "epoch": 0.5,
1825
+ "learning_rate": 0.00010132889178674283,
1826
+ "loss": 0.1187,
1827
+ "step": 303
1828
+ },
1829
+ {
1830
+ "epoch": 0.51,
1831
+ "learning_rate": 0.00010079735009246167,
1832
+ "loss": 0.1162,
1833
+ "step": 304
1834
+ },
1835
+ {
1836
+ "epoch": 0.51,
1837
+ "learning_rate": 0.00010026578586757778,
1838
+ "loss": 0.1128,
1839
+ "step": 305
1840
+ },
1841
+ {
1842
+ "epoch": 0.51,
1843
+ "learning_rate": 9.973421413242225e-05,
1844
+ "loss": 0.1187,
1845
+ "step": 306
1846
+ },
1847
+ {
1848
+ "epoch": 0.51,
1849
+ "learning_rate": 9.920264990753837e-05,
1850
+ "loss": 0.1238,
1851
+ "step": 307
1852
+ },
1853
+ {
1854
+ "epoch": 0.51,
1855
+ "learning_rate": 9.867110821325717e-05,
1856
+ "loss": 0.128,
1857
+ "step": 308
1858
+ },
1859
+ {
1860
+ "epoch": 0.51,
1861
+ "learning_rate": 9.813960406927319e-05,
1862
+ "loss": 0.1277,
1863
+ "step": 309
1864
+ },
1865
+ {
1866
+ "epoch": 0.52,
1867
+ "learning_rate": 9.760815249421973e-05,
1868
+ "loss": 0.1131,
1869
+ "step": 310
1870
+ },
1871
+ {
1872
+ "epoch": 0.52,
1873
+ "learning_rate": 9.707676850524473e-05,
1874
+ "loss": 0.116,
1875
+ "step": 311
1876
+ },
1877
+ {
1878
+ "epoch": 0.52,
1879
+ "learning_rate": 9.654546711758645e-05,
1880
+ "loss": 0.1076,
1881
+ "step": 312
1882
+ },
1883
+ {
1884
+ "epoch": 0.52,
1885
+ "learning_rate": 9.601426334414898e-05,
1886
+ "loss": 0.1117,
1887
+ "step": 313
1888
+ },
1889
+ {
1890
+ "epoch": 0.52,
1891
+ "learning_rate": 9.548317219507815e-05,
1892
+ "loss": 0.1418,
1893
+ "step": 314
1894
+ },
1895
+ {
1896
+ "epoch": 0.52,
1897
+ "learning_rate": 9.495220867733738e-05,
1898
+ "loss": 0.1216,
1899
+ "step": 315
1900
+ },
1901
+ {
1902
+ "epoch": 0.53,
1903
+ "learning_rate": 9.442138779428376e-05,
1904
+ "loss": 0.1156,
1905
+ "step": 316
1906
+ },
1907
+ {
1908
+ "epoch": 0.53,
1909
+ "learning_rate": 9.38907245452438e-05,
1910
+ "loss": 0.1208,
1911
+ "step": 317
1912
+ },
1913
+ {
1914
+ "epoch": 0.53,
1915
+ "learning_rate": 9.33602339250898e-05,
1916
+ "loss": 0.1244,
1917
+ "step": 318
1918
+ },
1919
+ {
1920
+ "epoch": 0.53,
1921
+ "learning_rate": 9.282993092381625e-05,
1922
+ "loss": 0.124,
1923
+ "step": 319
1924
+ },
1925
+ {
1926
+ "epoch": 0.53,
1927
+ "learning_rate": 9.229983052611597e-05,
1928
+ "loss": 0.1151,
1929
+ "step": 320
1930
+ },
1931
+ {
1932
+ "epoch": 0.53,
1933
+ "learning_rate": 9.176994771095687e-05,
1934
+ "loss": 0.1076,
1935
+ "step": 321
1936
+ },
1937
+ {
1938
+ "epoch": 0.54,
1939
+ "learning_rate": 9.12402974511587e-05,
1940
+ "loss": 0.1269,
1941
+ "step": 322
1942
+ },
1943
+ {
1944
+ "epoch": 0.54,
1945
+ "learning_rate": 9.071089471296995e-05,
1946
+ "loss": 0.1077,
1947
+ "step": 323
1948
+ },
1949
+ {
1950
+ "epoch": 0.54,
1951
+ "learning_rate": 9.018175445564485e-05,
1952
+ "loss": 0.1229,
1953
+ "step": 324
1954
+ },
1955
+ {
1956
+ "epoch": 0.54,
1957
+ "learning_rate": 8.965289163102078e-05,
1958
+ "loss": 0.1107,
1959
+ "step": 325
1960
+ },
1961
+ {
1962
+ "epoch": 0.54,
1963
+ "learning_rate": 8.912432118309582e-05,
1964
+ "loss": 0.1057,
1965
+ "step": 326
1966
+ },
1967
+ {
1968
+ "epoch": 0.54,
1969
+ "learning_rate": 8.859605804760626e-05,
1970
+ "loss": 0.107,
1971
+ "step": 327
1972
+ },
1973
+ {
1974
+ "epoch": 0.55,
1975
+ "learning_rate": 8.806811715160485e-05,
1976
+ "loss": 0.1142,
1977
+ "step": 328
1978
+ },
1979
+ {
1980
+ "epoch": 0.55,
1981
+ "learning_rate": 8.754051341303875e-05,
1982
+ "loss": 0.1188,
1983
+ "step": 329
1984
+ },
1985
+ {
1986
+ "epoch": 0.55,
1987
+ "learning_rate": 8.70132617403282e-05,
1988
+ "loss": 0.1011,
1989
+ "step": 330
1990
+ },
1991
+ {
1992
+ "epoch": 0.55,
1993
+ "learning_rate": 8.648637703194516e-05,
1994
+ "loss": 0.1168,
1995
+ "step": 331
1996
+ },
1997
+ {
1998
+ "epoch": 0.55,
1999
+ "learning_rate": 8.595987417599225e-05,
2000
+ "loss": 0.1239,
2001
+ "step": 332
2002
+ },
2003
+ {
2004
+ "epoch": 0.55,
2005
+ "learning_rate": 8.543376804978224e-05,
2006
+ "loss": 0.1039,
2007
+ "step": 333
2008
+ },
2009
+ {
2010
+ "epoch": 0.56,
2011
+ "learning_rate": 8.490807351941753e-05,
2012
+ "loss": 0.1113,
2013
+ "step": 334
2014
+ },
2015
+ {
2016
+ "epoch": 0.56,
2017
+ "learning_rate": 8.438280543937011e-05,
2018
+ "loss": 0.1179,
2019
+ "step": 335
2020
+ },
2021
+ {
2022
+ "epoch": 0.56,
2023
+ "learning_rate": 8.385797865206178e-05,
2024
+ "loss": 0.1257,
2025
+ "step": 336
2026
+ },
2027
+ {
2028
+ "epoch": 0.56,
2029
+ "learning_rate": 8.333360798744496e-05,
2030
+ "loss": 0.1193,
2031
+ "step": 337
2032
+ },
2033
+ {
2034
+ "epoch": 0.56,
2035
+ "learning_rate": 8.280970826258329e-05,
2036
+ "loss": 0.1265,
2037
+ "step": 338
2038
+ },
2039
+ {
2040
+ "epoch": 0.56,
2041
+ "learning_rate": 8.228629428123319e-05,
2042
+ "loss": 0.1181,
2043
+ "step": 339
2044
+ },
2045
+ {
2046
+ "epoch": 0.57,
2047
+ "learning_rate": 8.176338083342561e-05,
2048
+ "loss": 0.113,
2049
+ "step": 340
2050
+ },
2051
+ {
2052
+ "epoch": 0.57,
2053
+ "learning_rate": 8.124098269504787e-05,
2054
+ "loss": 0.1159,
2055
+ "step": 341
2056
+ },
2057
+ {
2058
+ "epoch": 0.57,
2059
+ "learning_rate": 8.07191146274263e-05,
2060
+ "loss": 0.1301,
2061
+ "step": 342
2062
+ },
2063
+ {
2064
+ "epoch": 0.57,
2065
+ "learning_rate": 8.019779137690906e-05,
2066
+ "loss": 0.1203,
2067
+ "step": 343
2068
+ },
2069
+ {
2070
+ "epoch": 0.57,
2071
+ "learning_rate": 7.967702767444964e-05,
2072
+ "loss": 0.1164,
2073
+ "step": 344
2074
+ },
2075
+ {
2076
+ "epoch": 0.57,
2077
+ "learning_rate": 7.915683823519031e-05,
2078
+ "loss": 0.1198,
2079
+ "step": 345
2080
+ },
2081
+ {
2082
+ "epoch": 0.58,
2083
+ "learning_rate": 7.863723775804651e-05,
2084
+ "loss": 0.1027,
2085
+ "step": 346
2086
+ },
2087
+ {
2088
+ "epoch": 0.58,
2089
+ "learning_rate": 7.811824092529156e-05,
2090
+ "loss": 0.1379,
2091
+ "step": 347
2092
+ },
2093
+ {
2094
+ "epoch": 0.58,
2095
+ "learning_rate": 7.759986240214155e-05,
2096
+ "loss": 0.1112,
2097
+ "step": 348
2098
+ },
2099
+ {
2100
+ "epoch": 0.58,
2101
+ "learning_rate": 7.708211683634112e-05,
2102
+ "loss": 0.1413,
2103
+ "step": 349
2104
+ },
2105
+ {
2106
+ "epoch": 0.58,
2107
+ "learning_rate": 7.656501885774964e-05,
2108
+ "loss": 0.1247,
2109
+ "step": 350
2110
+ },
2111
+ {
2112
+ "epoch": 0.58,
2113
+ "learning_rate": 7.604858307792758e-05,
2114
+ "loss": 0.1168,
2115
+ "step": 351
2116
+ },
2117
+ {
2118
+ "epoch": 0.59,
2119
+ "learning_rate": 7.553282408972382e-05,
2120
+ "loss": 0.1218,
2121
+ "step": 352
2122
+ },
2123
+ {
2124
+ "epoch": 0.59,
2125
+ "learning_rate": 7.501775646686315e-05,
2126
+ "loss": 0.1177,
2127
+ "step": 353
2128
+ },
2129
+ {
2130
+ "epoch": 0.59,
2131
+ "learning_rate": 7.450339476353474e-05,
2132
+ "loss": 0.1244,
2133
+ "step": 354
2134
+ },
2135
+ {
2136
+ "epoch": 0.59,
2137
+ "learning_rate": 7.398975351398053e-05,
2138
+ "loss": 0.1132,
2139
+ "step": 355
2140
+ },
2141
+ {
2142
+ "epoch": 0.59,
2143
+ "learning_rate": 7.34768472320847e-05,
2144
+ "loss": 0.1257,
2145
+ "step": 356
2146
+ },
2147
+ {
2148
+ "epoch": 0.59,
2149
+ "learning_rate": 7.29646904109637e-05,
2150
+ "loss": 0.1247,
2151
+ "step": 357
2152
+ },
2153
+ {
2154
+ "epoch": 0.6,
2155
+ "learning_rate": 7.245329752255647e-05,
2156
+ "loss": 0.1196,
2157
+ "step": 358
2158
+ },
2159
+ {
2160
+ "epoch": 0.6,
2161
+ "learning_rate": 7.194268301721563e-05,
2162
+ "loss": 0.1139,
2163
+ "step": 359
2164
+ },
2165
+ {
2166
+ "epoch": 0.6,
2167
+ "learning_rate": 7.143286132329912e-05,
2168
+ "loss": 0.129,
2169
+ "step": 360
2170
+ },
2171
+ {
2172
+ "epoch": 0.6,
2173
+ "learning_rate": 7.092384684676262e-05,
2174
+ "loss": 0.1164,
2175
+ "step": 361
2176
+ },
2177
+ {
2178
+ "epoch": 0.6,
2179
+ "learning_rate": 7.041565397075232e-05,
2180
+ "loss": 0.1048,
2181
+ "step": 362
2182
+ },
2183
+ {
2184
+ "epoch": 0.6,
2185
+ "learning_rate": 6.990829705519852e-05,
2186
+ "loss": 0.1098,
2187
+ "step": 363
2188
+ },
2189
+ {
2190
+ "epoch": 0.61,
2191
+ "learning_rate": 6.940179043641005e-05,
2192
+ "loss": 0.1129,
2193
+ "step": 364
2194
+ },
2195
+ {
2196
+ "epoch": 0.61,
2197
+ "learning_rate": 6.889614842666892e-05,
2198
+ "loss": 0.1162,
2199
+ "step": 365
2200
+ },
2201
+ {
2202
+ "epoch": 0.61,
2203
+ "learning_rate": 6.839138531382603e-05,
2204
+ "loss": 0.1198,
2205
+ "step": 366
2206
+ },
2207
+ {
2208
+ "epoch": 0.61,
2209
+ "learning_rate": 6.788751536089739e-05,
2210
+ "loss": 0.1107,
2211
+ "step": 367
2212
+ },
2213
+ {
2214
+ "epoch": 0.61,
2215
+ "learning_rate": 6.738455280566124e-05,
2216
+ "loss": 0.1177,
2217
+ "step": 368
2218
+ },
2219
+ {
2220
+ "epoch": 0.61,
2221
+ "learning_rate": 6.68825118602555e-05,
2222
+ "loss": 0.1189,
2223
+ "step": 369
2224
+ },
2225
+ {
2226
+ "epoch": 0.62,
2227
+ "learning_rate": 6.638140671077633e-05,
2228
+ "loss": 0.1105,
2229
+ "step": 370
2230
+ },
2231
+ {
2232
+ "epoch": 0.62,
2233
+ "learning_rate": 6.58812515168773e-05,
2234
+ "loss": 0.1156,
2235
+ "step": 371
2236
+ },
2237
+ {
2238
+ "epoch": 0.62,
2239
+ "learning_rate": 6.538206041136915e-05,
2240
+ "loss": 0.1306,
2241
+ "step": 372
2242
+ },
2243
+ {
2244
+ "epoch": 0.62,
2245
+ "learning_rate": 6.488384749982053e-05,
2246
+ "loss": 0.1218,
2247
+ "step": 373
2248
+ },
2249
+ {
2250
+ "epoch": 0.62,
2251
+ "learning_rate": 6.438662686015947e-05,
2252
+ "loss": 0.1162,
2253
+ "step": 374
2254
+ },
2255
+ {
2256
+ "epoch": 0.62,
2257
+ "learning_rate": 6.389041254227547e-05,
2258
+ "loss": 0.0903,
2259
+ "step": 375
2260
+ },
2261
+ {
2262
+ "epoch": 0.63,
2263
+ "learning_rate": 6.339521856762254e-05,
2264
+ "loss": 0.106,
2265
+ "step": 376
2266
+ },
2267
+ {
2268
+ "epoch": 0.63,
2269
+ "learning_rate": 6.290105892882303e-05,
2270
+ "loss": 0.1147,
2271
+ "step": 377
2272
+ },
2273
+ {
2274
+ "epoch": 0.63,
2275
+ "learning_rate": 6.240794758927221e-05,
2276
+ "loss": 0.1274,
2277
+ "step": 378
2278
+ },
2279
+ {
2280
+ "epoch": 0.63,
2281
+ "learning_rate": 6.191589848274368e-05,
2282
+ "loss": 0.1024,
2283
+ "step": 379
2284
+ },
2285
+ {
2286
+ "epoch": 0.63,
2287
+ "learning_rate": 6.142492551299576e-05,
2288
+ "loss": 0.1056,
2289
+ "step": 380
2290
+ },
2291
+ {
2292
+ "epoch": 0.63,
2293
+ "learning_rate": 6.093504255337844e-05,
2294
+ "loss": 0.1185,
2295
+ "step": 381
2296
+ },
2297
+ {
2298
+ "epoch": 0.64,
2299
+ "learning_rate": 6.044626344644151e-05,
2300
+ "loss": 0.1125,
2301
+ "step": 382
2302
+ },
2303
+ {
2304
+ "epoch": 0.64,
2305
+ "learning_rate": 5.995860200354335e-05,
2306
+ "loss": 0.118,
2307
+ "step": 383
2308
+ },
2309
+ {
2310
+ "epoch": 0.64,
2311
+ "learning_rate": 5.9472072004460665e-05,
2312
+ "loss": 0.1193,
2313
+ "step": 384
2314
+ },
2315
+ {
2316
+ "epoch": 0.64,
2317
+ "learning_rate": 5.8986687196999135e-05,
2318
+ "loss": 0.1221,
2319
+ "step": 385
2320
+ },
2321
+ {
2322
+ "epoch": 0.64,
2323
+ "learning_rate": 5.8502461296604935e-05,
2324
+ "loss": 0.1215,
2325
+ "step": 386
2326
+ },
2327
+ {
2328
+ "epoch": 0.64,
2329
+ "learning_rate": 5.801940798597716e-05,
2330
+ "loss": 0.1191,
2331
+ "step": 387
2332
+ },
2333
+ {
2334
+ "epoch": 0.65,
2335
+ "learning_rate": 5.753754091468115e-05,
2336
+ "loss": 0.1258,
2337
+ "step": 388
2338
+ },
2339
+ {
2340
+ "epoch": 0.65,
2341
+ "learning_rate": 5.7056873698763034e-05,
2342
+ "loss": 0.1149,
2343
+ "step": 389
2344
+ },
2345
+ {
2346
+ "epoch": 0.65,
2347
+ "learning_rate": 5.6577419920364625e-05,
2348
+ "loss": 0.1275,
2349
+ "step": 390
2350
+ },
2351
+ {
2352
+ "epoch": 0.65,
2353
+ "learning_rate": 5.6099193127339864e-05,
2354
+ "loss": 0.1203,
2355
+ "step": 391
2356
+ },
2357
+ {
2358
+ "epoch": 0.65,
2359
+ "learning_rate": 5.562220683287205e-05,
2360
+ "loss": 0.1165,
2361
+ "step": 392
2362
+ },
2363
+ {
2364
+ "epoch": 0.65,
2365
+ "learning_rate": 5.5146474515091754e-05,
2366
+ "loss": 0.1248,
2367
+ "step": 393
2368
+ },
2369
+ {
2370
+ "epoch": 0.66,
2371
+ "learning_rate": 5.467200961669619e-05,
2372
+ "loss": 0.0989,
2373
+ "step": 394
2374
+ },
2375
+ {
2376
+ "epoch": 0.66,
2377
+ "learning_rate": 5.4198825544569234e-05,
2378
+ "loss": 0.1127,
2379
+ "step": 395
2380
+ },
2381
+ {
2382
+ "epoch": 0.66,
2383
+ "learning_rate": 5.372693566940277e-05,
2384
+ "loss": 0.1248,
2385
+ "step": 396
2386
+ },
2387
+ {
2388
+ "epoch": 0.66,
2389
+ "learning_rate": 5.325635332531864e-05,
2390
+ "loss": 0.1094,
2391
+ "step": 397
2392
+ },
2393
+ {
2394
+ "epoch": 0.66,
2395
+ "learning_rate": 5.278709180949195e-05,
2396
+ "loss": 0.1194,
2397
+ "step": 398
2398
+ },
2399
+ {
2400
+ "epoch": 0.66,
2401
+ "learning_rate": 5.2319164381775524e-05,
2402
+ "loss": 0.1178,
2403
+ "step": 399
2404
+ },
2405
+ {
2406
+ "epoch": 0.67,
2407
+ "learning_rate": 5.1852584264324866e-05,
2408
+ "loss": 0.123,
2409
+ "step": 400
2410
+ },
2411
+ {
2412
+ "epoch": 0.67,
2413
+ "learning_rate": 5.138736464122484e-05,
2414
+ "loss": 0.1102,
2415
+ "step": 401
2416
+ },
2417
+ {
2418
+ "epoch": 0.67,
2419
+ "learning_rate": 5.092351865811698e-05,
2420
+ "loss": 0.1215,
2421
+ "step": 402
2422
+ },
2423
+ {
2424
+ "epoch": 0.67,
2425
+ "learning_rate": 5.046105942182815e-05,
2426
+ "loss": 0.115,
2427
+ "step": 403
2428
+ },
2429
+ {
2430
+ "epoch": 0.67,
2431
+ "learning_rate": 5.000000000000002e-05,
2432
+ "loss": 0.1181,
2433
+ "step": 404
2434
+ },
2435
+ {
2436
+ "epoch": 0.67,
2437
+ "learning_rate": 4.9540353420719946e-05,
2438
+ "loss": 0.096,
2439
+ "step": 405
2440
+ },
2441
+ {
2442
+ "epoch": 0.68,
2443
+ "learning_rate": 4.908213267215287e-05,
2444
+ "loss": 0.1176,
2445
+ "step": 406
2446
+ },
2447
+ {
2448
+ "epoch": 0.68,
2449
+ "learning_rate": 4.8625350702174166e-05,
2450
+ "loss": 0.1229,
2451
+ "step": 407
2452
+ },
2453
+ {
2454
+ "epoch": 0.68,
2455
+ "learning_rate": 4.817002041800388e-05,
2456
+ "loss": 0.1177,
2457
+ "step": 408
2458
+ },
2459
+ {
2460
+ "epoch": 0.68,
2461
+ "learning_rate": 4.7716154685841944e-05,
2462
+ "loss": 0.1124,
2463
+ "step": 409
2464
+ },
2465
+ {
2466
+ "epoch": 0.68,
2467
+ "learning_rate": 4.726376633050479e-05,
2468
+ "loss": 0.1325,
2469
+ "step": 410
2470
+ },
2471
+ {
2472
+ "epoch": 0.68,
2473
+ "learning_rate": 4.68128681350627e-05,
2474
+ "loss": 0.1072,
2475
+ "step": 411
2476
+ },
2477
+ {
2478
+ "epoch": 0.69,
2479
+ "learning_rate": 4.636347284047877e-05,
2480
+ "loss": 0.1168,
2481
+ "step": 412
2482
+ },
2483
+ {
2484
+ "epoch": 0.69,
2485
+ "learning_rate": 4.5915593145248924e-05,
2486
+ "loss": 0.105,
2487
+ "step": 413
2488
+ },
2489
+ {
2490
+ "epoch": 0.69,
2491
+ "learning_rate": 4.546924170504292e-05,
2492
+ "loss": 0.1121,
2493
+ "step": 414
2494
+ },
2495
+ {
2496
+ "epoch": 0.69,
2497
+ "learning_rate": 4.502443113234688e-05,
2498
+ "loss": 0.1075,
2499
+ "step": 415
2500
+ },
2501
+ {
2502
+ "epoch": 0.69,
2503
+ "learning_rate": 4.4581173996106815e-05,
2504
+ "loss": 0.1266,
2505
+ "step": 416
2506
+ },
2507
+ {
2508
+ "epoch": 0.69,
2509
+ "learning_rate": 4.413948282137367e-05,
2510
+ "loss": 0.1161,
2511
+ "step": 417
2512
+ },
2513
+ {
2514
+ "epoch": 0.7,
2515
+ "learning_rate": 4.3699370088949066e-05,
2516
+ "loss": 0.114,
2517
+ "step": 418
2518
+ },
2519
+ {
2520
+ "epoch": 0.7,
2521
+ "learning_rate": 4.326084823503287e-05,
2522
+ "loss": 0.1145,
2523
+ "step": 419
2524
+ },
2525
+ {
2526
+ "epoch": 0.7,
2527
+ "learning_rate": 4.282392965087182e-05,
2528
+ "loss": 0.1174,
2529
+ "step": 420
2530
+ },
2531
+ {
2532
+ "epoch": 0.7,
2533
+ "learning_rate": 4.2388626682409194e-05,
2534
+ "loss": 0.1368,
2535
+ "step": 421
2536
+ },
2537
+ {
2538
+ "epoch": 0.7,
2539
+ "learning_rate": 4.1954951629936065e-05,
2540
+ "loss": 0.1133,
2541
+ "step": 422
2542
+ },
2543
+ {
2544
+ "epoch": 0.7,
2545
+ "learning_rate": 4.152291674774383e-05,
2546
+ "loss": 0.1073,
2547
+ "step": 423
2548
+ },
2549
+ {
2550
+ "epoch": 0.71,
2551
+ "learning_rate": 4.109253424377772e-05,
2552
+ "loss": 0.0974,
2553
+ "step": 424
2554
+ },
2555
+ {
2556
+ "epoch": 0.71,
2557
+ "learning_rate": 4.0663816279292024e-05,
2558
+ "loss": 0.1427,
2559
+ "step": 425
2560
+ },
2561
+ {
2562
+ "epoch": 0.71,
2563
+ "learning_rate": 4.02367749685063e-05,
2564
+ "loss": 0.1349,
2565
+ "step": 426
2566
+ },
2567
+ {
2568
+ "epoch": 0.71,
2569
+ "learning_rate": 3.981142237826332e-05,
2570
+ "loss": 0.118,
2571
+ "step": 427
2572
+ },
2573
+ {
2574
+ "epoch": 0.71,
2575
+ "learning_rate": 3.93877705276878e-05,
2576
+ "loss": 0.1023,
2577
+ "step": 428
2578
+ },
2579
+ {
2580
+ "epoch": 0.71,
2581
+ "learning_rate": 3.896583138784688e-05,
2582
+ "loss": 0.0998,
2583
+ "step": 429
2584
+ },
2585
+ {
2586
+ "epoch": 0.72,
2587
+ "learning_rate": 3.854561688141205e-05,
2588
+ "loss": 0.112,
2589
+ "step": 430
2590
+ },
2591
+ {
2592
+ "epoch": 0.72,
2593
+ "learning_rate": 3.812713888232193e-05,
2594
+ "loss": 0.12,
2595
+ "step": 431
2596
+ },
2597
+ {
2598
+ "epoch": 0.72,
2599
+ "learning_rate": 3.7710409215446986e-05,
2600
+ "loss": 0.1135,
2601
+ "step": 432
2602
+ },
2603
+ {
2604
+ "epoch": 0.72,
2605
+ "learning_rate": 3.729543965625526e-05,
2606
+ "loss": 0.1196,
2607
+ "step": 433
2608
+ },
2609
+ {
2610
+ "epoch": 0.72,
2611
+ "learning_rate": 3.6882241930479824e-05,
2612
+ "loss": 0.1124,
2613
+ "step": 434
2614
+ },
2615
+ {
2616
+ "epoch": 0.72,
2617
+ "learning_rate": 3.6470827713787194e-05,
2618
+ "loss": 0.1165,
2619
+ "step": 435
2620
+ },
2621
+ {
2622
+ "epoch": 0.73,
2623
+ "learning_rate": 3.606120863144753e-05,
2624
+ "loss": 0.1031,
2625
+ "step": 436
2626
+ },
2627
+ {
2628
+ "epoch": 0.73,
2629
+ "learning_rate": 3.5653396258006265e-05,
2630
+ "loss": 0.1115,
2631
+ "step": 437
2632
+ },
2633
+ {
2634
+ "epoch": 0.73,
2635
+ "learning_rate": 3.524740211695683e-05,
2636
+ "loss": 0.1261,
2637
+ "step": 438
2638
+ },
2639
+ {
2640
+ "epoch": 0.73,
2641
+ "learning_rate": 3.4843237680415156e-05,
2642
+ "loss": 0.1213,
2643
+ "step": 439
2644
+ },
2645
+ {
2646
+ "epoch": 0.73,
2647
+ "learning_rate": 3.444091436879545e-05,
2648
+ "loss": 0.1236,
2649
+ "step": 440
2650
+ },
2651
+ {
2652
+ "epoch": 0.73,
2653
+ "learning_rate": 3.4040443550487645e-05,
2654
+ "loss": 0.1157,
2655
+ "step": 441
2656
+ },
2657
+ {
2658
+ "epoch": 0.74,
2659
+ "learning_rate": 3.364183654153592e-05,
2660
+ "loss": 0.1072,
2661
+ "step": 442
2662
+ },
2663
+ {
2664
+ "epoch": 0.74,
2665
+ "learning_rate": 3.32451046053191e-05,
2666
+ "loss": 0.0938,
2667
+ "step": 443
2668
+ },
2669
+ {
2670
+ "epoch": 0.74,
2671
+ "learning_rate": 3.285025895223244e-05,
2672
+ "loss": 0.1236,
2673
+ "step": 444
2674
+ },
2675
+ {
2676
+ "epoch": 0.74,
2677
+ "learning_rate": 3.245731073937068e-05,
2678
+ "loss": 0.1369,
2679
+ "step": 445
2680
+ },
2681
+ {
2682
+ "epoch": 0.74,
2683
+ "learning_rate": 3.2066271070212874e-05,
2684
+ "loss": 0.1168,
2685
+ "step": 446
2686
+ },
2687
+ {
2688
+ "epoch": 0.74,
2689
+ "learning_rate": 3.167715099430873e-05,
2690
+ "loss": 0.0973,
2691
+ "step": 447
2692
+ },
2693
+ {
2694
+ "epoch": 0.75,
2695
+ "learning_rate": 3.1289961506966214e-05,
2696
+ "loss": 0.0985,
2697
+ "step": 448
2698
+ },
2699
+ {
2700
+ "epoch": 0.75,
2701
+ "learning_rate": 3.0904713548940936e-05,
2702
+ "loss": 0.1168,
2703
+ "step": 449
2704
+ },
2705
+ {
2706
+ "epoch": 0.75,
2707
+ "learning_rate": 3.052141800612709e-05,
2708
+ "loss": 0.1115,
2709
+ "step": 450
2710
+ },
2711
+ {
2712
+ "epoch": 0.75,
2713
+ "learning_rate": 3.0140085709249667e-05,
2714
+ "loss": 0.1155,
2715
+ "step": 451
2716
+ },
2717
+ {
2718
+ "epoch": 0.75,
2719
+ "learning_rate": 2.9760727433558522e-05,
2720
+ "loss": 0.1223,
2721
+ "step": 452
2722
+ },
2723
+ {
2724
+ "epoch": 0.75,
2725
+ "learning_rate": 2.938335389852397e-05,
2726
+ "loss": 0.1252,
2727
+ "step": 453
2728
+ },
2729
+ {
2730
+ "epoch": 0.76,
2731
+ "learning_rate": 2.9007975767533714e-05,
2732
+ "loss": 0.1272,
2733
+ "step": 454
2734
+ },
2735
+ {
2736
+ "epoch": 0.76,
2737
+ "learning_rate": 2.863460364759163e-05,
2738
+ "loss": 0.114,
2739
+ "step": 455
2740
+ },
2741
+ {
2742
+ "epoch": 0.76,
2743
+ "learning_rate": 2.8263248089018113e-05,
2744
+ "loss": 0.0958,
2745
+ "step": 456
2746
+ },
2747
+ {
2748
+ "epoch": 0.76,
2749
+ "learning_rate": 2.789391958515183e-05,
2750
+ "loss": 0.1294,
2751
+ "step": 457
2752
+ },
2753
+ {
2754
+ "epoch": 0.76,
2755
+ "learning_rate": 2.7526628572053227e-05,
2756
+ "loss": 0.1126,
2757
+ "step": 458
2758
+ },
2759
+ {
2760
+ "epoch": 0.76,
2761
+ "learning_rate": 2.7161385428209774e-05,
2762
+ "loss": 0.11,
2763
+ "step": 459
2764
+ },
2765
+ {
2766
+ "epoch": 0.77,
2767
+ "learning_rate": 2.679820047424253e-05,
2768
+ "loss": 0.1189,
2769
+ "step": 460
2770
+ },
2771
+ {
2772
+ "epoch": 0.77,
2773
+ "learning_rate": 2.6437083972614572e-05,
2774
+ "loss": 0.1115,
2775
+ "step": 461
2776
+ },
2777
+ {
2778
+ "epoch": 0.77,
2779
+ "learning_rate": 2.6078046127341137e-05,
2780
+ "loss": 0.1205,
2781
+ "step": 462
2782
+ },
2783
+ {
2784
+ "epoch": 0.77,
2785
+ "learning_rate": 2.5721097083701084e-05,
2786
+ "loss": 0.096,
2787
+ "step": 463
2788
+ },
2789
+ {
2790
+ "epoch": 0.77,
2791
+ "learning_rate": 2.5366246927950286e-05,
2792
+ "loss": 0.1064,
2793
+ "step": 464
2794
+ },
2795
+ {
2796
+ "epoch": 0.77,
2797
+ "learning_rate": 2.5013505687036786e-05,
2798
+ "loss": 0.1279,
2799
+ "step": 465
2800
+ },
2801
+ {
2802
+ "epoch": 0.78,
2803
+ "learning_rate": 2.4662883328317222e-05,
2804
+ "loss": 0.1124,
2805
+ "step": 466
2806
+ },
2807
+ {
2808
+ "epoch": 0.78,
2809
+ "learning_rate": 2.4314389759275335e-05,
2810
+ "loss": 0.1052,
2811
+ "step": 467
2812
+ },
2813
+ {
2814
+ "epoch": 0.78,
2815
+ "learning_rate": 2.3968034827241925e-05,
2816
+ "loss": 0.1103,
2817
+ "step": 468
2818
+ },
2819
+ {
2820
+ "epoch": 0.78,
2821
+ "learning_rate": 2.3623828319116748e-05,
2822
+ "loss": 0.1075,
2823
+ "step": 469
2824
+ },
2825
+ {
2826
+ "epoch": 0.78,
2827
+ "learning_rate": 2.3281779961091775e-05,
2828
+ "loss": 0.0996,
2829
+ "step": 470
2830
+ },
2831
+ {
2832
+ "epoch": 0.78,
2833
+ "learning_rate": 2.2941899418376466e-05,
2834
+ "loss": 0.0935,
2835
+ "step": 471
2836
+ },
2837
+ {
2838
+ "epoch": 0.79,
2839
+ "learning_rate": 2.2604196294924694e-05,
2840
+ "loss": 0.1096,
2841
+ "step": 472
2842
+ },
2843
+ {
2844
+ "epoch": 0.79,
2845
+ "learning_rate": 2.2268680133163277e-05,
2846
+ "loss": 0.1101,
2847
+ "step": 473
2848
+ },
2849
+ {
2850
+ "epoch": 0.79,
2851
+ "learning_rate": 2.1935360413722395e-05,
2852
+ "loss": 0.1223,
2853
+ "step": 474
2854
+ },
2855
+ {
2856
+ "epoch": 0.79,
2857
+ "learning_rate": 2.1604246555167638e-05,
2858
+ "loss": 0.1165,
2859
+ "step": 475
2860
+ },
2861
+ {
2862
+ "epoch": 0.79,
2863
+ "learning_rate": 2.1275347913734022e-05,
2864
+ "loss": 0.1182,
2865
+ "step": 476
2866
+ },
2867
+ {
2868
+ "epoch": 0.79,
2869
+ "learning_rate": 2.0948673783061422e-05,
2870
+ "loss": 0.1039,
2871
+ "step": 477
2872
+ },
2873
+ {
2874
+ "epoch": 0.8,
2875
+ "learning_rate": 2.0624233393932024e-05,
2876
+ "loss": 0.1117,
2877
+ "step": 478
2878
+ },
2879
+ {
2880
+ "epoch": 0.8,
2881
+ "learning_rate": 2.03020359140096e-05,
2882
+ "loss": 0.1178,
2883
+ "step": 479
2884
+ },
2885
+ {
2886
+ "epoch": 0.8,
2887
+ "learning_rate": 1.9982090447580303e-05,
2888
+ "loss": 0.1253,
2889
+ "step": 480
2890
+ },
2891
+ {
2892
+ "epoch": 0.8,
2893
+ "learning_rate": 1.966440603529549e-05,
2894
+ "loss": 0.1079,
2895
+ "step": 481
2896
+ },
2897
+ {
2898
+ "epoch": 0.8,
2899
+ "learning_rate": 1.9348991653916228e-05,
2900
+ "loss": 0.101,
2901
+ "step": 482
2902
+ },
2903
+ {
2904
+ "epoch": 0.8,
2905
+ "learning_rate": 1.9035856216059722e-05,
2906
+ "loss": 0.1012,
2907
+ "step": 483
2908
+ },
2909
+ {
2910
+ "epoch": 0.81,
2911
+ "learning_rate": 1.8725008569947365e-05,
2912
+ "loss": 0.121,
2913
+ "step": 484
2914
+ },
2915
+ {
2916
+ "epoch": 0.81,
2917
+ "learning_rate": 1.8416457499154728e-05,
2918
+ "loss": 0.1232,
2919
+ "step": 485
2920
+ },
2921
+ {
2922
+ "epoch": 0.81,
2923
+ "learning_rate": 1.811021172236348e-05,
2924
+ "loss": 0.1058,
2925
+ "step": 486
2926
+ },
2927
+ {
2928
+ "epoch": 0.81,
2929
+ "learning_rate": 1.7806279893114875e-05,
2930
+ "loss": 0.1078,
2931
+ "step": 487
2932
+ },
2933
+ {
2934
+ "epoch": 0.81,
2935
+ "learning_rate": 1.750467059956531e-05,
2936
+ "loss": 0.1188,
2937
+ "step": 488
2938
+ },
2939
+ {
2940
+ "epoch": 0.81,
2941
+ "learning_rate": 1.7205392364243623e-05,
2942
+ "loss": 0.118,
2943
+ "step": 489
2944
+ },
2945
+ {
2946
+ "epoch": 0.82,
2947
+ "learning_rate": 1.690845364381034e-05,
2948
+ "loss": 0.108,
2949
+ "step": 490
2950
+ },
2951
+ {
2952
+ "epoch": 0.82,
2953
+ "learning_rate": 1.6613862828818628e-05,
2954
+ "loss": 0.1035,
2955
+ "step": 491
2956
+ },
2957
+ {
2958
+ "epoch": 0.82,
2959
+ "learning_rate": 1.6321628243477194e-05,
2960
+ "loss": 0.1274,
2961
+ "step": 492
2962
+ },
2963
+ {
2964
+ "epoch": 0.82,
2965
+ "learning_rate": 1.603175814541522e-05,
2966
+ "loss": 0.1184,
2967
+ "step": 493
2968
+ },
2969
+ {
2970
+ "epoch": 0.82,
2971
+ "learning_rate": 1.5744260725448844e-05,
2972
+ "loss": 0.1038,
2973
+ "step": 494
2974
+ },
2975
+ {
2976
+ "epoch": 0.82,
2977
+ "learning_rate": 1.5459144107349787e-05,
2978
+ "loss": 0.1059,
2979
+ "step": 495
2980
+ },
2981
+ {
2982
+ "epoch": 0.83,
2983
+ "learning_rate": 1.5176416347615885e-05,
2984
+ "loss": 0.1079,
2985
+ "step": 496
2986
+ },
2987
+ {
2988
+ "epoch": 0.83,
2989
+ "learning_rate": 1.4896085435243279e-05,
2990
+ "loss": 0.1169,
2991
+ "step": 497
2992
+ },
2993
+ {
2994
+ "epoch": 0.83,
2995
+ "learning_rate": 1.4618159291500778e-05,
2996
+ "loss": 0.1092,
2997
+ "step": 498
2998
+ },
2999
+ {
3000
+ "epoch": 0.83,
3001
+ "learning_rate": 1.4342645769705977e-05,
3002
+ "loss": 0.1362,
3003
+ "step": 499
3004
+ },
3005
+ {
3006
+ "epoch": 0.83,
3007
+ "learning_rate": 1.406955265500346e-05,
3008
+ "loss": 0.1393,
3009
+ "step": 500
3010
+ },
3011
+ {
3012
+ "epoch": 0.83,
3013
+ "learning_rate": 1.3798887664144633e-05,
3014
+ "loss": 0.1252,
3015
+ "step": 501
3016
+ },
3017
+ {
3018
+ "epoch": 0.84,
3019
+ "learning_rate": 1.3530658445269783e-05,
3020
+ "loss": 0.1019,
3021
+ "step": 502
3022
+ },
3023
+ {
3024
+ "epoch": 0.84,
3025
+ "learning_rate": 1.3264872577692022e-05,
3026
+ "loss": 0.1204,
3027
+ "step": 503
3028
+ },
3029
+ {
3030
+ "epoch": 0.84,
3031
+ "learning_rate": 1.3001537571682965e-05,
3032
+ "loss": 0.1303,
3033
+ "step": 504
3034
+ },
3035
+ {
3036
+ "epoch": 0.84,
3037
+ "learning_rate": 1.2740660868260633e-05,
3038
+ "loss": 0.1178,
3039
+ "step": 505
3040
+ },
3041
+ {
3042
+ "epoch": 0.84,
3043
+ "learning_rate": 1.2482249838979142e-05,
3044
+ "loss": 0.1066,
3045
+ "step": 506
3046
+ },
3047
+ {
3048
+ "epoch": 0.84,
3049
+ "learning_rate": 1.2226311785720468e-05,
3050
+ "loss": 0.1184,
3051
+ "step": 507
3052
+ },
3053
+ {
3054
+ "epoch": 0.85,
3055
+ "learning_rate": 1.1972853940488015e-05,
3056
+ "loss": 0.1062,
3057
+ "step": 508
3058
+ },
3059
+ {
3060
+ "epoch": 0.85,
3061
+ "learning_rate": 1.1721883465202332e-05,
3062
+ "loss": 0.1105,
3063
+ "step": 509
3064
+ },
3065
+ {
3066
+ "epoch": 0.85,
3067
+ "learning_rate": 1.1473407451498753e-05,
3068
+ "loss": 0.1232,
3069
+ "step": 510
3070
+ },
3071
+ {
3072
+ "epoch": 0.85,
3073
+ "learning_rate": 1.122743292052697e-05,
3074
+ "loss": 0.1225,
3075
+ "step": 511
3076
+ },
3077
+ {
3078
+ "epoch": 0.85,
3079
+ "learning_rate": 1.0983966822752623e-05,
3080
+ "loss": 0.1021,
3081
+ "step": 512
3082
+ },
3083
+ {
3084
+ "epoch": 0.85,
3085
+ "learning_rate": 1.0743016037760945e-05,
3086
+ "loss": 0.1062,
3087
+ "step": 513
3088
+ },
3089
+ {
3090
+ "epoch": 0.86,
3091
+ "learning_rate": 1.0504587374062391e-05,
3092
+ "loss": 0.116,
3093
+ "step": 514
3094
+ },
3095
+ {
3096
+ "epoch": 0.86,
3097
+ "learning_rate": 1.026868756890016e-05,
3098
+ "loss": 0.1203,
3099
+ "step": 515
3100
+ },
3101
+ {
3102
+ "epoch": 0.86,
3103
+ "learning_rate": 1.003532328805986e-05,
3104
+ "loss": 0.117,
3105
+ "step": 516
3106
+ },
3107
+ {
3108
+ "epoch": 0.86,
3109
+ "learning_rate": 9.804501125681243e-06,
3110
+ "loss": 0.1043,
3111
+ "step": 517
3112
+ },
3113
+ {
3114
+ "epoch": 0.86,
3115
+ "learning_rate": 9.57622760407173e-06,
3116
+ "loss": 0.1162,
3117
+ "step": 518
3118
+ },
3119
+ {
3120
+ "epoch": 0.86,
3121
+ "learning_rate": 9.350509173522193e-06,
3122
+ "loss": 0.1154,
3123
+ "step": 519
3124
+ },
3125
+ {
3126
+ "epoch": 0.87,
3127
+ "learning_rate": 9.127352212124662e-06,
3128
+ "loss": 0.1088,
3129
+ "step": 520
3130
+ },
3131
+ {
3132
+ "epoch": 0.87,
3133
+ "learning_rate": 8.90676302559219e-06,
3134
+ "loss": 0.1187,
3135
+ "step": 521
3136
+ },
3137
+ {
3138
+ "epoch": 0.87,
3139
+ "learning_rate": 8.688747847080514e-06,
3140
+ "loss": 0.1084,
3141
+ "step": 522
3142
+ },
3143
+ {
3144
+ "epoch": 0.87,
3145
+ "learning_rate": 8.473312837012026e-06,
3146
+ "loss": 0.1071,
3147
+ "step": 523
3148
+ },
3149
+ {
3150
+ "epoch": 0.87,
3151
+ "learning_rate": 8.260464082901732e-06,
3152
+ "loss": 0.1241,
3153
+ "step": 524
3154
+ },
3155
+ {
3156
+ "epoch": 0.87,
3157
+ "learning_rate": 8.050207599185134e-06,
3158
+ "loss": 0.1227,
3159
+ "step": 525
3160
+ },
3161
+ {
3162
+ "epoch": 0.88,
3163
+ "learning_rate": 7.842549327048365e-06,
3164
+ "loss": 0.1189,
3165
+ "step": 526
3166
+ },
3167
+ {
3168
+ "epoch": 0.88,
3169
+ "learning_rate": 7.637495134260242e-06,
3170
+ "loss": 0.1025,
3171
+ "step": 527
3172
+ },
3173
+ {
3174
+ "epoch": 0.88,
3175
+ "learning_rate": 7.435050815006561e-06,
3176
+ "loss": 0.1024,
3177
+ "step": 528
3178
+ },
3179
+ {
3180
+ "epoch": 0.88,
3181
+ "learning_rate": 7.235222089726279e-06,
3182
+ "loss": 0.0958,
3183
+ "step": 529
3184
+ },
3185
+ {
3186
+ "epoch": 0.88,
3187
+ "learning_rate": 7.038014604949883e-06,
3188
+ "loss": 0.1155,
3189
+ "step": 530
3190
+ },
3191
+ {
3192
+ "epoch": 0.88,
3193
+ "learning_rate": 6.843433933139909e-06,
3194
+ "loss": 0.1107,
3195
+ "step": 531
3196
+ },
3197
+ {
3198
+ "epoch": 0.89,
3199
+ "learning_rate": 6.651485572533378e-06,
3200
+ "loss": 0.1109,
3201
+ "step": 532
3202
+ },
3203
+ {
3204
+ "epoch": 0.89,
3205
+ "learning_rate": 6.46217494698651e-06,
3206
+ "loss": 0.1211,
3207
+ "step": 533
3208
+ },
3209
+ {
3210
+ "epoch": 0.89,
3211
+ "learning_rate": 6.275507405821435e-06,
3212
+ "loss": 0.121,
3213
+ "step": 534
3214
+ },
3215
+ {
3216
+ "epoch": 0.89,
3217
+ "learning_rate": 6.091488223675057e-06,
3218
+ "loss": 0.114,
3219
+ "step": 535
3220
+ },
3221
+ {
3222
+ "epoch": 0.89,
3223
+ "learning_rate": 5.910122600349965e-06,
3224
+ "loss": 0.1285,
3225
+ "step": 536
3226
+ },
3227
+ {
3228
+ "epoch": 0.89,
3229
+ "learning_rate": 5.7314156606675496e-06,
3230
+ "loss": 0.0938,
3231
+ "step": 537
3232
+ },
3233
+ {
3234
+ "epoch": 0.9,
3235
+ "learning_rate": 5.5553724543231825e-06,
3236
+ "loss": 0.1107,
3237
+ "step": 538
3238
+ },
3239
+ {
3240
+ "epoch": 0.9,
3241
+ "learning_rate": 5.381997955743501e-06,
3242
+ "loss": 0.1255,
3243
+ "step": 539
3244
+ },
3245
+ {
3246
+ "epoch": 0.9,
3247
+ "learning_rate": 5.2112970639458745e-06,
3248
+ "loss": 0.1278,
3249
+ "step": 540
3250
+ },
3251
+ {
3252
+ "epoch": 0.9,
3253
+ "learning_rate": 5.043274602399939e-06,
3254
+ "loss": 0.1134,
3255
+ "step": 541
3256
+ },
3257
+ {
3258
+ "epoch": 0.9,
3259
+ "learning_rate": 4.87793531889138e-06,
3260
+ "loss": 0.1072,
3261
+ "step": 542
3262
+ },
3263
+ {
3264
+ "epoch": 0.9,
3265
+ "learning_rate": 4.715283885387678e-06,
3266
+ "loss": 0.1115,
3267
+ "step": 543
3268
+ },
3269
+ {
3270
+ "epoch": 0.91,
3271
+ "learning_rate": 4.555324897906132e-06,
3272
+ "loss": 0.1012,
3273
+ "step": 544
3274
+ },
3275
+ {
3276
+ "epoch": 0.91,
3277
+ "learning_rate": 4.398062876384046e-06,
3278
+ "loss": 0.1145,
3279
+ "step": 545
3280
+ },
3281
+ {
3282
+ "epoch": 0.91,
3283
+ "learning_rate": 4.2435022645509025e-06,
3284
+ "loss": 0.1079,
3285
+ "step": 546
3286
+ },
3287
+ {
3288
+ "epoch": 0.91,
3289
+ "learning_rate": 4.091647429802869e-06,
3290
+ "loss": 0.1133,
3291
+ "step": 547
3292
+ },
3293
+ {
3294
+ "epoch": 0.91,
3295
+ "learning_rate": 3.942502663079395e-06,
3296
+ "loss": 0.1078,
3297
+ "step": 548
3298
+ },
3299
+ {
3300
+ "epoch": 0.91,
3301
+ "learning_rate": 3.796072178741916e-06,
3302
+ "loss": 0.1113,
3303
+ "step": 549
3304
+ },
3305
+ {
3306
+ "epoch": 0.92,
3307
+ "learning_rate": 3.6523601144548003e-06,
3308
+ "loss": 3.1026,
3309
+ "step": 550
3310
+ },
3311
+ {
3312
+ "epoch": 0.92,
3313
+ "learning_rate": 3.5113705310684363e-06,
3314
+ "loss": 0.1238,
3315
+ "step": 551
3316
+ },
3317
+ {
3318
+ "epoch": 0.92,
3319
+ "learning_rate": 3.3731074125044726e-06,
3320
+ "loss": 0.1266,
3321
+ "step": 552
3322
+ },
3323
+ {
3324
+ "epoch": 0.92,
3325
+ "learning_rate": 3.2375746656432284e-06,
3326
+ "loss": 0.1344,
3327
+ "step": 553
3328
+ },
3329
+ {
3330
+ "epoch": 0.92,
3331
+ "learning_rate": 3.1047761202133597e-06,
3332
+ "loss": 0.1211,
3333
+ "step": 554
3334
+ },
3335
+ {
3336
+ "epoch": 0.92,
3337
+ "learning_rate": 2.974715528683547e-06,
3338
+ "loss": 0.1223,
3339
+ "step": 555
3340
+ },
3341
+ {
3342
+ "epoch": 0.93,
3343
+ "learning_rate": 2.8473965661565347e-06,
3344
+ "loss": 0.1148,
3345
+ "step": 556
3346
+ },
3347
+ {
3348
+ "epoch": 0.93,
3349
+ "learning_rate": 2.7228228302653034e-06,
3350
+ "loss": 0.1238,
3351
+ "step": 557
3352
+ },
3353
+ {
3354
+ "epoch": 0.93,
3355
+ "learning_rate": 2.600997841071329e-06,
3356
+ "loss": 0.0977,
3357
+ "step": 558
3358
+ },
3359
+ {
3360
+ "epoch": 0.93,
3361
+ "learning_rate": 2.4819250409651607e-06,
3362
+ "loss": 0.1203,
3363
+ "step": 559
3364
+ },
3365
+ {
3366
+ "epoch": 0.93,
3367
+ "learning_rate": 2.3656077945691803e-06,
3368
+ "loss": 0.1029,
3369
+ "step": 560
3370
+ },
3371
+ {
3372
+ "epoch": 0.93,
3373
+ "learning_rate": 2.2520493886424743e-06,
3374
+ "loss": 0.1305,
3375
+ "step": 561
3376
+ },
3377
+ {
3378
+ "epoch": 0.94,
3379
+ "learning_rate": 2.1412530319879887e-06,
3380
+ "loss": 0.1208,
3381
+ "step": 562
3382
+ },
3383
+ {
3384
+ "epoch": 0.94,
3385
+ "learning_rate": 2.0332218553618885e-06,
3386
+ "loss": 0.1234,
3387
+ "step": 563
3388
+ },
3389
+ {
3390
+ "epoch": 0.94,
3391
+ "learning_rate": 1.9279589113850084e-06,
3392
+ "loss": 0.1387,
3393
+ "step": 564
3394
+ },
3395
+ {
3396
+ "epoch": 0.94,
3397
+ "learning_rate": 1.825467174456652e-06,
3398
+ "loss": 0.1313,
3399
+ "step": 565
3400
+ },
3401
+ {
3402
+ "epoch": 0.94,
3403
+ "learning_rate": 1.725749540670596e-06,
3404
+ "loss": 0.1135,
3405
+ "step": 566
3406
+ },
3407
+ {
3408
+ "epoch": 0.94,
3409
+ "learning_rate": 1.6288088277331304e-06,
3410
+ "loss": 0.101,
3411
+ "step": 567
3412
+ },
3413
+ {
3414
+ "epoch": 0.95,
3415
+ "learning_rate": 1.5346477748835354e-06,
3416
+ "loss": 0.1131,
3417
+ "step": 568
3418
+ },
3419
+ {
3420
+ "epoch": 0.95,
3421
+ "learning_rate": 1.4432690428166528e-06,
3422
+ "loss": 0.1117,
3423
+ "step": 569
3424
+ },
3425
+ {
3426
+ "epoch": 0.95,
3427
+ "learning_rate": 1.3546752136076923e-06,
3428
+ "loss": 0.1107,
3429
+ "step": 570
3430
+ },
3431
+ {
3432
+ "epoch": 0.95,
3433
+ "learning_rate": 1.268868790639277e-06,
3434
+ "loss": 0.115,
3435
+ "step": 571
3436
+ },
3437
+ {
3438
+ "epoch": 0.95,
3439
+ "learning_rate": 1.1858521985307125e-06,
3440
+ "loss": 0.1163,
3441
+ "step": 572
3442
+ },
3443
+ {
3444
+ "epoch": 0.95,
3445
+ "learning_rate": 1.105627783069485e-06,
3446
+ "loss": 0.1131,
3447
+ "step": 573
3448
+ },
3449
+ {
3450
+ "epoch": 0.96,
3451
+ "learning_rate": 1.0281978111449375e-06,
3452
+ "loss": 0.1085,
3453
+ "step": 574
3454
+ },
3455
+ {
3456
+ "epoch": 0.96,
3457
+ "learning_rate": 9.535644706842317e-07,
3458
+ "loss": 0.1257,
3459
+ "step": 575
3460
+ },
3461
+ {
3462
+ "epoch": 0.96,
3463
+ "learning_rate": 8.817298705905641e-07,
3464
+ "loss": 0.1188,
3465
+ "step": 576
3466
+ },
3467
+ {
3468
+ "epoch": 0.96,
3469
+ "learning_rate": 8.126960406835249e-07,
3470
+ "loss": 0.1101,
3471
+ "step": 577
3472
+ },
3473
+ {
3474
+ "epoch": 0.96,
3475
+ "learning_rate": 7.464649316417438e-07,
3476
+ "loss": 0.1075,
3477
+ "step": 578
3478
+ },
3479
+ {
3480
+ "epoch": 0.96,
3481
+ "learning_rate": 6.830384149478008e-07,
3482
+ "loss": 0.0966,
3483
+ "step": 579
3484
+ },
3485
+ {
3486
+ "epoch": 0.97,
3487
+ "learning_rate": 6.224182828353242e-07,
3488
+ "loss": 0.1244,
3489
+ "step": 580
3490
+ },
3491
+ {
3492
+ "epoch": 0.97,
3493
+ "learning_rate": 5.64606248238364e-07,
3494
+ "loss": 0.1137,
3495
+ "step": 581
3496
+ },
3497
+ {
3498
+ "epoch": 0.97,
3499
+ "learning_rate": 5.096039447429534e-07,
3500
+ "loss": 0.1087,
3501
+ "step": 582
3502
+ },
3503
+ {
3504
+ "epoch": 0.97,
3505
+ "learning_rate": 4.57412926541001e-07,
3506
+ "loss": 0.1022,
3507
+ "step": 583
3508
+ },
3509
+ {
3510
+ "epoch": 0.97,
3511
+ "learning_rate": 4.0803466838631455e-07,
3512
+ "loss": 0.1104,
3513
+ "step": 584
3514
+ },
3515
+ {
3516
+ "epoch": 0.97,
3517
+ "learning_rate": 3.614705655529682e-07,
3518
+ "loss": 0.1169,
3519
+ "step": 585
3520
+ },
3521
+ {
3522
+ "epoch": 0.98,
3523
+ "learning_rate": 3.177219337958892e-07,
3524
+ "loss": 0.1176,
3525
+ "step": 586
3526
+ },
3527
+ {
3528
+ "epoch": 0.98,
3529
+ "learning_rate": 2.767900093136544e-07,
3530
+ "loss": 0.1273,
3531
+ "step": 587
3532
+ },
3533
+ {
3534
+ "epoch": 0.98,
3535
+ "learning_rate": 2.3867594871352926e-07,
3536
+ "loss": 0.1032,
3537
+ "step": 588
3538
+ },
3539
+ {
3540
+ "epoch": 0.98,
3541
+ "learning_rate": 2.0338082897886079e-07,
3542
+ "loss": 0.1342,
3543
+ "step": 589
3544
+ },
3545
+ {
3546
+ "epoch": 0.98,
3547
+ "learning_rate": 1.709056474385795e-07,
3548
+ "loss": 0.1124,
3549
+ "step": 590
3550
+ },
3551
+ {
3552
+ "epoch": 0.98,
3553
+ "learning_rate": 1.412513217390554e-07,
3554
+ "loss": 0.1231,
3555
+ "step": 591
3556
+ },
3557
+ {
3558
+ "epoch": 0.99,
3559
+ "learning_rate": 1.1441868981815207e-07,
3560
+ "loss": 0.1221,
3561
+ "step": 592
3562
+ },
3563
+ {
3564
+ "epoch": 0.99,
3565
+ "learning_rate": 9.040850988153438e-08,
3566
+ "loss": 0.1088,
3567
+ "step": 593
3568
+ },
3569
+ {
3570
+ "epoch": 0.99,
3571
+ "learning_rate": 6.922146038129684e-08,
3572
+ "loss": 0.1155,
3573
+ "step": 594
3574
+ },
3575
+ {
3576
+ "epoch": 0.99,
3577
+ "learning_rate": 5.08581399967345e-08,
3578
+ "loss": 0.1103,
3579
+ "step": 595
3580
+ },
3581
+ {
3582
+ "epoch": 0.99,
3583
+ "learning_rate": 3.53190676174453e-08,
3584
+ "loss": 0.1189,
3585
+ "step": 596
3586
+ },
3587
+ {
3588
+ "epoch": 0.99,
3589
+ "learning_rate": 2.260468232869739e-08,
3590
+ "loss": 0.1202,
3591
+ "step": 597
3592
+ },
3593
+ {
3594
+ "epoch": 1.0,
3595
+ "learning_rate": 1.2715343398972402e-08,
3596
+ "loss": 0.1187,
3597
+ "step": 598
3598
+ },
3599
+ {
3600
+ "epoch": 1.0,
3601
+ "learning_rate": 5.651330269840216e-09,
3602
+ "loss": 0.0985,
3603
+ "step": 599
3604
+ },
3605
+ {
3606
+ "epoch": 1.0,
3607
+ "learning_rate": 1.4128425480763874e-09,
3608
+ "loss": 0.1111,
3609
+ "step": 600
3610
+ },
3611
+ {
3612
+ "epoch": 1.0,
3613
+ "learning_rate": 0.0,
3614
+ "loss": 0.1056,
3615
+ "step": 601
3616
+ }
3617
+ ],
3618
+ "logging_steps": 1,
3619
+ "max_steps": 601,
3620
+ "num_input_tokens_seen": 0,
3621
+ "num_train_epochs": 1,
3622
+ "save_steps": 500,
3623
+ "total_flos": 188987860844544.0,
3624
+ "train_batch_size": 16,
3625
+ "trial_name": null,
3626
+ "trial_params": null
3627
+ }
checkpoint-601/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:df3b1605591803a5caaa97ea0ad69cb877d516fa0d9f22146e358860e06bbd46
3
+ size 6200
checkpoint-601/zero_to_fp32.py ADDED
@@ -0,0 +1,592 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/usr/bin/env python
2
+
3
+ # Copyright (c) Microsoft Corporation.
4
+ # SPDX-License-Identifier: Apache-2.0
5
+
6
+ # DeepSpeed Team
7
+
8
+ # This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
9
+ # copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
10
+ # the future. Once extracted, the weights don't require DeepSpeed and can be used in any
11
+ # application.
12
+ #
13
+ # example: python zero_to_fp32.py . pytorch_model.bin
14
+
15
+ import argparse
16
+ import torch
17
+ import glob
18
+ import math
19
+ import os
20
+ import re
21
+ from collections import OrderedDict
22
+ from dataclasses import dataclass
23
+
24
+ # while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
25
+ # DeepSpeed data structures it has to be available in the current python environment.
26
+ from deepspeed.utils import logger
27
+ from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
28
+ FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
29
+ FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
30
+
31
+
32
+ @dataclass
33
+ class zero_model_state:
34
+ buffers: dict()
35
+ param_shapes: dict()
36
+ shared_params: list
37
+ ds_version: int
38
+ frozen_param_shapes: dict()
39
+ frozen_param_fragments: dict()
40
+
41
+
42
+ debug = 0
43
+
44
+ # load to cpu
45
+ device = torch.device('cpu')
46
+
47
+
48
+ def atoi(text):
49
+ return int(text) if text.isdigit() else text
50
+
51
+
52
+ def natural_keys(text):
53
+ '''
54
+ alist.sort(key=natural_keys) sorts in human order
55
+ http://nedbatchelder.com/blog/200712/human_sorting.html
56
+ (See Toothy's implementation in the comments)
57
+ '''
58
+ return [atoi(c) for c in re.split(r'(\d+)', text)]
59
+
60
+
61
+ def get_model_state_file(checkpoint_dir, zero_stage):
62
+ if not os.path.isdir(checkpoint_dir):
63
+ raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
64
+
65
+ # there should be only one file
66
+ if zero_stage <= 2:
67
+ file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
68
+ elif zero_stage == 3:
69
+ file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
70
+
71
+ if not os.path.exists(file):
72
+ raise FileNotFoundError(f"can't find model states file at '{file}'")
73
+
74
+ return file
75
+
76
+
77
+ def get_checkpoint_files(checkpoint_dir, glob_pattern):
78
+ # XXX: need to test that this simple glob rule works for multi-node setup too
79
+ ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
80
+
81
+ if len(ckpt_files) == 0:
82
+ raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
83
+
84
+ return ckpt_files
85
+
86
+
87
+ def get_optim_files(checkpoint_dir):
88
+ return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
89
+
90
+
91
+ def get_model_state_files(checkpoint_dir):
92
+ return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
93
+
94
+
95
+ def parse_model_states(files):
96
+ zero_model_states = []
97
+ for file in files:
98
+ state_dict = torch.load(file, map_location=device)
99
+
100
+ if BUFFER_NAMES not in state_dict:
101
+ raise ValueError(f"{file} is not a model state checkpoint")
102
+ buffer_names = state_dict[BUFFER_NAMES]
103
+ if debug:
104
+ print("Found buffers:", buffer_names)
105
+
106
+ # recover just the buffers while restoring them to fp32 if they were saved in fp16
107
+ buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
108
+ param_shapes = state_dict[PARAM_SHAPES]
109
+
110
+ # collect parameters that are included in param_shapes
111
+ param_names = []
112
+ for s in param_shapes:
113
+ for name in s.keys():
114
+ param_names.append(name)
115
+
116
+ # update with frozen parameters
117
+ frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
118
+ if frozen_param_shapes is not None:
119
+ if debug:
120
+ print(f"Found frozen_param_shapes: {frozen_param_shapes}")
121
+ param_names += list(frozen_param_shapes.keys())
122
+
123
+ # handle shared params
124
+ shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
125
+
126
+ ds_version = state_dict.get(DS_VERSION, None)
127
+
128
+ frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
129
+
130
+ z_model_state = zero_model_state(buffers=buffers,
131
+ param_shapes=param_shapes,
132
+ shared_params=shared_params,
133
+ ds_version=ds_version,
134
+ frozen_param_shapes=frozen_param_shapes,
135
+ frozen_param_fragments=frozen_param_fragments)
136
+ zero_model_states.append(z_model_state)
137
+
138
+ return zero_model_states
139
+
140
+
141
+ def parse_optim_states(files, ds_checkpoint_dir):
142
+
143
+ total_files = len(files)
144
+ state_dicts = []
145
+ for f in files:
146
+ state_dict = torch.load(f, map_location=device)
147
+ # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
148
+ # and also handle the case where it was already removed by another helper script
149
+ state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
150
+ state_dicts.append(state_dict)
151
+
152
+ if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
153
+ raise ValueError(f"{files[0]} is not a zero checkpoint")
154
+ zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
155
+ world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
156
+
157
+ # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
158
+ # parameters can be different from data parallelism for non-expert parameters. So we can just
159
+ # use the max of the partition_count to get the dp world_size.
160
+
161
+ if type(world_size) is list:
162
+ world_size = max(world_size)
163
+
164
+ if world_size != total_files:
165
+ raise ValueError(
166
+ f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
167
+ "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
168
+ )
169
+
170
+ # the groups are named differently in each stage
171
+ if zero_stage <= 2:
172
+ fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
173
+ elif zero_stage == 3:
174
+ fp32_groups_key = FP32_FLAT_GROUPS
175
+ else:
176
+ raise ValueError(f"unknown zero stage {zero_stage}")
177
+
178
+ if zero_stage <= 2:
179
+ fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
180
+ elif zero_stage == 3:
181
+ # if there is more than one param group, there will be multiple flattened tensors - one
182
+ # flattened tensor per group - for simplicity merge them into a single tensor
183
+ #
184
+ # XXX: could make the script more memory efficient for when there are multiple groups - it
185
+ # will require matching the sub-lists of param_shapes for each param group flattened tensor
186
+
187
+ fp32_flat_groups = [
188
+ torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
189
+ ]
190
+
191
+ return zero_stage, world_size, fp32_flat_groups
192
+
193
+
194
+ def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir):
195
+ """
196
+ Returns fp32 state_dict reconstructed from ds checkpoint
197
+
198
+ Args:
199
+ - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
200
+
201
+ """
202
+ print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
203
+
204
+ optim_files = get_optim_files(ds_checkpoint_dir)
205
+ zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
206
+ print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
207
+
208
+ model_files = get_model_state_files(ds_checkpoint_dir)
209
+
210
+ zero_model_states = parse_model_states(model_files)
211
+ print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
212
+
213
+ if zero_stage <= 2:
214
+ return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states)
215
+ elif zero_stage == 3:
216
+ return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states)
217
+
218
+
219
+ def _zero2_merge_frozen_params(state_dict, zero_model_states):
220
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
221
+ return
222
+
223
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
224
+ frozen_param_fragments = zero_model_states[0].frozen_param_fragments
225
+
226
+ if debug:
227
+ num_elem = sum(s.numel() for s in frozen_param_shapes.values())
228
+ print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
229
+
230
+ wanted_params = len(frozen_param_shapes)
231
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
232
+ avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
233
+ print(f'Frozen params: Have {avail_numel} numels to process.')
234
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
235
+
236
+ total_params = 0
237
+ total_numel = 0
238
+ for name, shape in frozen_param_shapes.items():
239
+ total_params += 1
240
+ unpartitioned_numel = shape.numel()
241
+ total_numel += unpartitioned_numel
242
+
243
+ state_dict[name] = frozen_param_fragments[name]
244
+
245
+ if debug:
246
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
247
+
248
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
249
+
250
+
251
+ def _has_callable(obj, fn):
252
+ attr = getattr(obj, fn, None)
253
+ return callable(attr)
254
+
255
+
256
+ def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
257
+ param_shapes = zero_model_states[0].param_shapes
258
+
259
+ # Reconstruction protocol:
260
+ #
261
+ # XXX: document this
262
+
263
+ if debug:
264
+ for i in range(world_size):
265
+ for j in range(len(fp32_flat_groups[0])):
266
+ print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
267
+
268
+ # XXX: memory usage doubles here (zero2)
269
+ num_param_groups = len(fp32_flat_groups[0])
270
+ merged_single_partition_of_fp32_groups = []
271
+ for i in range(num_param_groups):
272
+ merged_partitions = [sd[i] for sd in fp32_flat_groups]
273
+ full_single_fp32_vector = torch.cat(merged_partitions, 0)
274
+ merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
275
+ avail_numel = sum(
276
+ [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
277
+
278
+ if debug:
279
+ wanted_params = sum([len(shapes) for shapes in param_shapes])
280
+ wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
281
+ # not asserting if there is a mismatch due to possible padding
282
+ print(f"Have {avail_numel} numels to process.")
283
+ print(f"Need {wanted_numel} numels in {wanted_params} params.")
284
+
285
+ # params
286
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
287
+ # out-of-core computing solution
288
+ total_numel = 0
289
+ total_params = 0
290
+ for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
291
+ offset = 0
292
+ avail_numel = full_single_fp32_vector.numel()
293
+ for name, shape in shapes.items():
294
+
295
+ unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
296
+ total_numel += unpartitioned_numel
297
+ total_params += 1
298
+
299
+ if debug:
300
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
301
+ state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
302
+ offset += unpartitioned_numel
303
+
304
+ # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
305
+ # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
306
+ # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
307
+ # live optimizer object, so we are checking that the numbers are within the right range
308
+ align_to = 2 * world_size
309
+
310
+ def zero2_align(x):
311
+ return align_to * math.ceil(x / align_to)
312
+
313
+ if debug:
314
+ print(f"original offset={offset}, avail_numel={avail_numel}")
315
+
316
+ offset = zero2_align(offset)
317
+ avail_numel = zero2_align(avail_numel)
318
+
319
+ if debug:
320
+ print(f"aligned offset={offset}, avail_numel={avail_numel}")
321
+
322
+ # Sanity check
323
+ if offset != avail_numel:
324
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
325
+
326
+ print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
327
+
328
+
329
+ def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states):
330
+ state_dict = OrderedDict()
331
+
332
+ # buffers
333
+ buffers = zero_model_states[0].buffers
334
+ state_dict.update(buffers)
335
+ if debug:
336
+ print(f"added {len(buffers)} buffers")
337
+
338
+ _zero2_merge_frozen_params(state_dict, zero_model_states)
339
+
340
+ _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
341
+
342
+ # recover shared parameters
343
+ for pair in zero_model_states[0].shared_params:
344
+ if pair[1] in state_dict:
345
+ state_dict[pair[0]] = state_dict[pair[1]]
346
+
347
+ return state_dict
348
+
349
+
350
+ def zero3_partitioned_param_info(unpartitioned_numel, world_size):
351
+ remainder = unpartitioned_numel % world_size
352
+ padding_numel = (world_size - remainder) if remainder else 0
353
+ partitioned_numel = math.ceil(unpartitioned_numel / world_size)
354
+ return partitioned_numel, padding_numel
355
+
356
+
357
+ def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
358
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
359
+ return
360
+
361
+ if debug:
362
+ for i in range(world_size):
363
+ num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
364
+ print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
365
+
366
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
367
+ wanted_params = len(frozen_param_shapes)
368
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
369
+ avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
370
+ print(f'Frozen params: Have {avail_numel} numels to process.')
371
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
372
+
373
+ total_params = 0
374
+ total_numel = 0
375
+ for name, shape in zero_model_states[0].frozen_param_shapes.items():
376
+ total_params += 1
377
+ unpartitioned_numel = shape.numel()
378
+ total_numel += unpartitioned_numel
379
+
380
+ param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
381
+ state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
382
+
383
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
384
+
385
+ if debug:
386
+ print(
387
+ f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
388
+ )
389
+
390
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
391
+
392
+
393
+ def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
394
+ param_shapes = zero_model_states[0].param_shapes
395
+ avail_numel = fp32_flat_groups[0].numel() * world_size
396
+ # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
397
+ # param, re-consolidating each param, while dealing with padding if any
398
+
399
+ # merge list of dicts, preserving order
400
+ param_shapes = {k: v for d in param_shapes for k, v in d.items()}
401
+
402
+ if debug:
403
+ for i in range(world_size):
404
+ print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
405
+
406
+ wanted_params = len(param_shapes)
407
+ wanted_numel = sum(shape.numel() for shape in param_shapes.values())
408
+ # not asserting if there is a mismatch due to possible padding
409
+ avail_numel = fp32_flat_groups[0].numel() * world_size
410
+ print(f"Trainable params: Have {avail_numel} numels to process.")
411
+ print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
412
+
413
+ # params
414
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
415
+ # out-of-core computing solution
416
+ offset = 0
417
+ total_numel = 0
418
+ total_params = 0
419
+ for name, shape in param_shapes.items():
420
+
421
+ unpartitioned_numel = shape.numel()
422
+ total_numel += unpartitioned_numel
423
+ total_params += 1
424
+
425
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
426
+
427
+ if debug:
428
+ print(
429
+ f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
430
+ )
431
+
432
+ # XXX: memory usage doubles here
433
+ state_dict[name] = torch.cat(
434
+ tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
435
+ 0).narrow(0, 0, unpartitioned_numel).view(shape)
436
+ offset += partitioned_numel
437
+
438
+ offset *= world_size
439
+
440
+ # Sanity check
441
+ if offset != avail_numel:
442
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
443
+
444
+ print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
445
+
446
+
447
+ def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states):
448
+ state_dict = OrderedDict()
449
+
450
+ # buffers
451
+ buffers = zero_model_states[0].buffers
452
+ state_dict.update(buffers)
453
+ if debug:
454
+ print(f"added {len(buffers)} buffers")
455
+
456
+ _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
457
+
458
+ _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
459
+
460
+ # recover shared parameters
461
+ for pair in zero_model_states[0].shared_params:
462
+ if pair[1] in state_dict:
463
+ state_dict[pair[0]] = state_dict[pair[1]]
464
+
465
+ return state_dict
466
+
467
+
468
+ def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None):
469
+ """
470
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
471
+ ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
472
+ via a model hub.
473
+
474
+ Args:
475
+ - ``checkpoint_dir``: path to the desired checkpoint folder
476
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
477
+
478
+ Returns:
479
+ - pytorch ``state_dict``
480
+
481
+ Note: this approach may not work if your application doesn't have sufficient free CPU memory and
482
+ you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
483
+ the checkpoint.
484
+
485
+ A typical usage might be ::
486
+
487
+ from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
488
+ # do the training and checkpoint saving
489
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
490
+ model = model.cpu() # move to cpu
491
+ model.load_state_dict(state_dict)
492
+ # submit to model hub or save the model to share with others
493
+
494
+ In this example the ``model`` will no longer be usable in the deepspeed context of the same
495
+ application. i.e. you will need to re-initialize the deepspeed engine, since
496
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
497
+
498
+ If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
499
+
500
+ """
501
+ if tag is None:
502
+ latest_path = os.path.join(checkpoint_dir, 'latest')
503
+ if os.path.isfile(latest_path):
504
+ with open(latest_path, 'r') as fd:
505
+ tag = fd.read().strip()
506
+ else:
507
+ raise ValueError(f"Unable to find 'latest' file at {latest_path}")
508
+
509
+ ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
510
+
511
+ if not os.path.isdir(ds_checkpoint_dir):
512
+ raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
513
+
514
+ return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir)
515
+
516
+
517
+ def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None):
518
+ """
519
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
520
+ loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
521
+
522
+ Args:
523
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
524
+ - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
525
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
526
+ """
527
+
528
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
529
+ print(f"Saving fp32 state dict to {output_file}")
530
+ torch.save(state_dict, output_file)
531
+
532
+
533
+ def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
534
+ """
535
+ 1. Put the provided model to cpu
536
+ 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
537
+ 3. Load it into the provided model
538
+
539
+ Args:
540
+ - ``model``: the model object to update
541
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
542
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
543
+
544
+ Returns:
545
+ - ``model`: modified model
546
+
547
+ Make sure you have plenty of CPU memory available before you call this function. If you don't
548
+ have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
549
+ conveniently placed for you in the checkpoint folder.
550
+
551
+ A typical usage might be ::
552
+
553
+ from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
554
+ model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
555
+ # submit to model hub or save the model to share with others
556
+
557
+ Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
558
+ of the same application. i.e. you will need to re-initialize the deepspeed engine, since
559
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
560
+
561
+ """
562
+ logger.info(f"Extracting fp32 weights")
563
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
564
+
565
+ logger.info(f"Overwriting model with fp32 weights")
566
+ model = model.cpu()
567
+ model.load_state_dict(state_dict, strict=False)
568
+
569
+ return model
570
+
571
+
572
+ if __name__ == "__main__":
573
+
574
+ parser = argparse.ArgumentParser()
575
+ parser.add_argument("checkpoint_dir",
576
+ type=str,
577
+ help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
578
+ parser.add_argument(
579
+ "output_file",
580
+ type=str,
581
+ help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
582
+ parser.add_argument("-t",
583
+ "--tag",
584
+ type=str,
585
+ default=None,
586
+ help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
587
+ parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
588
+ args = parser.parse_args()
589
+
590
+ debug = args.debug
591
+
592
+ convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, args.output_file, tag=args.tag)
config.json ADDED
@@ -0,0 +1,46 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "rizla/rizla-17",
3
+ "architectures": [
4
+ "MixtralForCausalLM"
5
+ ],
6
+ "attention_dropout": 0.0,
7
+ "bos_token_id": 1,
8
+ "eos_token_id": 2,
9
+ "hidden_act": "silu",
10
+ "hidden_size": 4096,
11
+ "initializer_range": 0.02,
12
+ "intermediate_size": 14336,
13
+ "max_position_embeddings": 32768,
14
+ "model_type": "mixtral",
15
+ "num_attention_heads": 32,
16
+ "num_experts_per_tok": 2,
17
+ "num_hidden_layers": 39,
18
+ "num_key_value_heads": 8,
19
+ "num_local_experts": 2,
20
+ "output_router_logits": true,
21
+ "pad_token_id": 2,
22
+ "quantization_config": {
23
+ "_load_in_4bit": true,
24
+ "_load_in_8bit": false,
25
+ "bnb_4bit_compute_dtype": "bfloat16",
26
+ "bnb_4bit_quant_type": "nf4",
27
+ "bnb_4bit_use_double_quant": true,
28
+ "llm_int8_enable_fp32_cpu_offload": false,
29
+ "llm_int8_has_fp16_weight": false,
30
+ "llm_int8_skip_modules": null,
31
+ "llm_int8_threshold": 6.0,
32
+ "load_in_4bit": true,
33
+ "load_in_8bit": false,
34
+ "quant_method": "bitsandbytes"
35
+ },
36
+ "rms_norm_eps": 1e-05,
37
+ "rope_theta": 10000.0,
38
+ "router_aux_loss_coef": 0.001,
39
+ "sliding_window": null,
40
+ "tie_word_embeddings": false,
41
+ "torch_dtype": "bfloat16",
42
+ "transformers_version": "4.38.0.dev0",
43
+ "unsloth_version": "2024.1",
44
+ "use_cache": false,
45
+ "vocab_size": 32000
46
+ }
runs/Feb03_03-36-39_19503a009d2c/events.out.tfevents.1706931401.19503a009d2c.4586.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ffb0948ed12228b1d57ad4cb2153b57236267e85037dd386b1729676a77e21a2
3
+ size 5564
runs/Feb03_03-38-23_19503a009d2c/events.out.tfevents.1706931504.19503a009d2c.4673.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e6ab7055cf7e2b4bae1ddef36c5591d58538c04237a2c96a71a93bb6bcddda58
3
+ size 5564
runs/Feb03_03-50-56_19503a009d2c/events.out.tfevents.1706932258.19503a009d2c.5617.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d37bae394faf91a963dc4209b0bc4057dc3742250103e6b5472256165a6a34b2
3
+ size 5870
runs/Feb03_03-53-37_19503a009d2c/events.out.tfevents.1706932424.19503a009d2c.5964.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:be4701fe827d4250b4fb4121db5276f2ef7d7bb352ef52325c1c530177c6ed93
3
+ size 5716
runs/Feb03_03-55-24_19503a009d2c/events.out.tfevents.1706932529.19503a009d2c.6106.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:98df81f9d07d7645ce7490868afb25c9c4bf99b95c3afbea0c58c416c878d3d6
3
+ size 6640
runs/Feb03_03-58-43_19503a009d2c/events.out.tfevents.1706932728.19503a009d2c.6196.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f06d9bb07b99f4d56f653df33c2d1f3c2ff2e082063c3f1fd47bd759e0632f1f
3
+ size 6794
runs/Feb03_04-04-38_19503a009d2c/events.out.tfevents.1706933083.19503a009d2c.6301.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cc4b76109d366c802c9c488c5bebe7a563cdc12655f38d020099b9950a77c7a6
3
+ size 5870
runs/Feb03_04-07-08_19503a009d2c/events.out.tfevents.1706933232.19503a009d2c.6407.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e9f0f43ca8c8b093875e5c522dd16ea13176b09067a0ba146a426c894ea8f747
3
+ size 6181
runs/Feb03_04-11-41_19503a009d2c/events.out.tfevents.1706933506.19503a009d2c.6498.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:20ecda04f64ab54300a9fa5e29b88e2083980a2b8dc82b40483f7c7f95c09aa2
3
+ size 5719
runs/Feb03_04-13-40_19503a009d2c/events.out.tfevents.1706933624.19503a009d2c.6813.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f2b156f8d5b7a39588ea49701e40dfddf8444c58fb2bc9c71930c7833bf363fd
3
+ size 99895
special_tokens_map.json ADDED
@@ -0,0 +1,35 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<unk>",
4
+ "<s>",
5
+ "</s>"
6
+ ],
7
+ "bos_token": {
8
+ "content": "<s>",
9
+ "lstrip": false,
10
+ "normalized": false,
11
+ "rstrip": false,
12
+ "single_word": false
13
+ },
14
+ "eos_token": {
15
+ "content": "</s>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false
20
+ },
21
+ "pad_token": {
22
+ "content": "<s>",
23
+ "lstrip": false,
24
+ "normalized": false,
25
+ "rstrip": false,
26
+ "single_word": false
27
+ },
28
+ "unk_token": {
29
+ "content": "<unk>",
30
+ "lstrip": false,
31
+ "normalized": false,
32
+ "rstrip": false,
33
+ "single_word": false
34
+ }
35
+ }
tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dadfd56d766715c61d2ef780a525ab43b8e6da4de6865bda3d95fdef5e134055
3
+ size 493443
tokenizer_config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "added_tokens_decoder": {
5
+ "0": {
6
+ "content": "<unk>",
7
+ "lstrip": false,
8
+ "normalized": false,
9
+ "rstrip": false,
10
+ "single_word": false,
11
+ "special": true
12
+ },
13
+ "1": {
14
+ "content": "<s>",
15
+ "lstrip": false,
16
+ "normalized": false,
17
+ "rstrip": false,
18
+ "single_word": false,
19
+ "special": true
20
+ },
21
+ "2": {
22
+ "content": "</s>",
23
+ "lstrip": false,
24
+ "normalized": false,
25
+ "rstrip": false,
26
+ "single_word": false,
27
+ "special": true
28
+ }
29
+ },
30
+ "additional_special_tokens": [
31
+ "<unk>",
32
+ "<s>",
33
+ "</s>"
34
+ ],
35
+ "bos_token": "<s>",
36
+ "clean_up_tokenization_spaces": false,
37
+ "eos_token": "</s>",
38
+ "legacy": true,
39
+ "max_length": null,
40
+ "model_max_length": 255,
41
+ "pad_to_multiple_of": null,
42
+ "pad_token": "<s>",
43
+ "pad_token_type_id": 0,
44
+ "padding_side": "left",
45
+ "sp_model_kwargs": {},
46
+ "spaces_between_special_tokens": false,
47
+ "tokenizer_class": "LlamaTokenizer",
48
+ "trust_remote_code": true,
49
+ "unk_token": "<unk>",
50
+ "use_default_system_prompt": true,
51
+ "use_fast": true
52
+ }