Upload 16 files

README.md

@@ -1,3 +1,202 @@
----
-license: apache-2.0
----
+---
+base_model: Qwen/Qwen2.5-7B
+library_name: peft
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- 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. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+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).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.13.2

adapter_config.json

@@ -0,0 +1,34 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 64,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "q_proj",
+    "o_proj",
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "k_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}

adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:78ecf58e6042ef5bb264de1efdb5a697a5f3d058311fee824b2f33e016044efe
+size 323014168

added_tokens.json

@@ -0,0 +1,24 @@
+{
+  "</tool_call>": 151658,
+  "<tool_call>": 151657,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

global_step3750/mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c3fe5d301fbdf6ea1a5a6a59bec25a8580ad325dbe3933e162572c1778ca7224
+size 328967600

global_step3750/zero_pp_rank_0_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:22cfcebfe678d3c231d351e85bfba20e327b9b18598bdf615a748533542893db
+size 1937815469

latest

@@ -0,0 +1 @@
+global_step3750

rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:98b76c8287641de7cc96e56274701d828e2a821d61b87e77cff3fb4cd172bc77
+size 14244

special_tokens_map.json

@@ -0,0 +1,25 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "<|endoftext|>"
+}

tokenizer_config.json

@@ -0,0 +1,209 @@
+{
+  "add_bos_token": false,
+  "add_eos_token": true,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "chat_template": "{%- if tools %}\n    {{- '<|im_start|>system\\n' }}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- messages[0]['content'] }}\n    {%- else %}\n        {{- 'You are a helpful assistant.' }}\n    {%- endif %}\n    {{- \"\\n\\n# Tools\\n\\nYou may call one or more functions to assist with the user query.\\n\\nYou are provided with function signatures within <tools></tools> XML tags:\\n<tools>\" }}\n    {%- for tool in tools %}\n        {{- \"\\n\" }}\n        {{- tool | tojson }}\n    {%- endfor %}\n    {{- \"\\n</tools>\\n\\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\\n<tool_call>\\n{\\\"name\\\": <function-name>, \\\"arguments\\\": <args-json-object>}\\n</tool_call><|im_end|>\\n\" }}\n{%- else %}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- '<|im_start|>system\\n' + messages[0]['content'] + '<|im_end|>\\n' }}\n    {%- else %}\n        {{- '<|im_start|>system\\nYou are a helpful assistant.<|im_end|>\\n' }}\n    {%- endif %}\n{%- endif %}\n{%- for message in messages %}\n    {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) or (message.role == \"assistant\" and not message.tool_calls) %}\n        {{- '<|im_start|>' + message.role + '\\n' + message.content + '<|im_end|>' + '\\n' }}\n    {%- elif message.role == \"assistant\" %}\n        {{- '<|im_start|>' + message.role }}\n        {%- if message.content %}\n            {{- '\\n' + message.content }}\n        {%- endif %}\n        {%- for tool_call in message.tool_calls %}\n            {%- if tool_call.function is defined %}\n                {%- set tool_call = tool_call.function %}\n            {%- endif %}\n            {{- '\\n<tool_call>\\n{\"name\": \"' }}\n            {{- tool_call.name }}\n            {{- '\", \"arguments\": ' }}\n            {{- tool_call.arguments | tojson }}\n            {{- '}\\n</tool_call>' }}\n        {%- endfor %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"tool\" %}\n        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != \"tool\") %}\n            {{- '<|im_start|>user' }}\n        {%- endif %}\n        {{- '\\n<tool_response>\\n' }}\n        {{- message.content }}\n        {{- '\\n</tool_response>' }}\n        {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n            {{- '<|im_end|>\\n' }}\n        {%- endif %}\n    {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n    {{- '<|im_start|>assistant\\n' }}\n{%- endif %}\n",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "padding_side": "left",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

trainer_state.json

@@ -0,0 +1,1542 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.07284240787863483,
+  "eval_steps": 75,
+  "global_step": 1875,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0003884928420193858,
+      "grad_norm": 0.49890583753585815,
+      "learning_rate": 7.759200756309999e-05,
+      "loss": 1.8971,
+      "step": 10
+    },
+    {
+      "epoch": 0.0007769856840387716,
+      "grad_norm": 1.6535608768463135,
+      "learning_rate": 0.0001,
+      "loss": 1.6661,
+      "step": 20
+    },
+    {
+      "epoch": 0.0011654785260581573,
+      "grad_norm": 1.0714327096939087,
+      "learning_rate": 0.0001,
+      "loss": 1.4684,
+      "step": 30
+    },
+    {
+      "epoch": 0.0015539713680775432,
+      "grad_norm": 0.6442272067070007,
+      "learning_rate": 0.0001,
+      "loss": 1.5553,
+      "step": 40
+    },
+    {
+      "epoch": 0.0019424642100969289,
+      "grad_norm": 0.818639874458313,
+      "learning_rate": 0.0001,
+      "loss": 1.5007,
+      "step": 50
+    },
+    {
+      "epoch": 0.0023309570521163146,
+      "grad_norm": 1.3463096618652344,
+      "learning_rate": 0.0001,
+      "loss": 1.2678,
+      "step": 60
+    },
+    {
+      "epoch": 0.0027194498941357005,
+      "grad_norm": 0.8409688472747803,
+      "learning_rate": 0.0001,
+      "loss": 1.4836,
+      "step": 70
+    },
+    {
+      "epoch": 0.0029136963151453936,
+      "eval_loss": 1.4955415725708008,
+      "eval_runtime": 328.1939,
+      "eval_samples_per_second": 1.587,
+      "eval_steps_per_second": 1.587,
+      "step": 75
+    },
+    {
+      "epoch": 0.0031079427361550864,
+      "grad_norm": 0.49311453104019165,
+      "learning_rate": 0.0001,
+      "loss": 1.5626,
+      "step": 80
+    },
+    {
+      "epoch": 0.0034964355781744723,
+      "grad_norm": 0.5505372881889343,
+      "learning_rate": 0.0001,
+      "loss": 1.4427,
+      "step": 90
+    },
+    {
+      "epoch": 0.0038849284201938577,
+      "grad_norm": 1.1296964883804321,
+      "learning_rate": 0.0001,
+      "loss": 1.4702,
+      "step": 100
+    },
+    {
+      "epoch": 0.004273421262213244,
+      "grad_norm": 1.38261878490448,
+      "learning_rate": 0.0001,
+      "loss": 1.5806,
+      "step": 110
+    },
+    {
+      "epoch": 0.004661914104232629,
+      "grad_norm": 0.5213516354560852,
+      "learning_rate": 0.0001,
+      "loss": 1.4675,
+      "step": 120
+    },
+    {
+      "epoch": 0.0050504069462520155,
+      "grad_norm": 2.93784761428833,
+      "learning_rate": 0.0001,
+      "loss": 1.6413,
+      "step": 130
+    },
+    {
+      "epoch": 0.005438899788271401,
+      "grad_norm": 0.6772735118865967,
+      "learning_rate": 0.0001,
+      "loss": 1.2942,
+      "step": 140
+    },
+    {
+      "epoch": 0.005827392630290787,
+      "grad_norm": 1.1066700220108032,
+      "learning_rate": 0.0001,
+      "loss": 1.4082,
+      "step": 150
+    },
+    {
+      "epoch": 0.005827392630290787,
+      "eval_loss": 1.4717903137207031,
+      "eval_runtime": 419.3037,
+      "eval_samples_per_second": 1.243,
+      "eval_steps_per_second": 1.243,
+      "step": 150
+    },
+    {
+      "epoch": 0.006215885472310173,
+      "grad_norm": 0.5077598690986633,
+      "learning_rate": 0.0001,
+      "loss": 1.6572,
+      "step": 160
+    },
+    {
+      "epoch": 0.006604378314329558,
+      "grad_norm": 0.5666481256484985,
+      "learning_rate": 0.0001,
+      "loss": 1.7344,
+      "step": 170
+    },
+    {
+      "epoch": 0.0069928711563489445,
+      "grad_norm": 0.7042965888977051,
+      "learning_rate": 0.0001,
+      "loss": 1.3634,
+      "step": 180
+    },
+    {
+      "epoch": 0.00738136399836833,
+      "grad_norm": 0.6379776000976562,
+      "learning_rate": 0.0001,
+      "loss": 1.6191,
+      "step": 190
+    },
+    {
+      "epoch": 0.0077698568403877155,
+      "grad_norm": 0.7309342622756958,
+      "learning_rate": 0.0001,
+      "loss": 1.2186,
+      "step": 200
+    },
+    {
+      "epoch": 0.008158349682407102,
+      "grad_norm": 1.4138643741607666,
+      "learning_rate": 0.0001,
+      "loss": 1.5201,
+      "step": 210
+    },
+    {
+      "epoch": 0.008546842524426487,
+      "grad_norm": 1.3856728076934814,
+      "learning_rate": 0.0001,
+      "loss": 1.4056,
+      "step": 220
+    },
+    {
+      "epoch": 0.00874108894543618,
+      "eval_loss": 1.473658800125122,
+      "eval_runtime": 413.7612,
+      "eval_samples_per_second": 1.259,
+      "eval_steps_per_second": 1.259,
+      "step": 225
+    },
+    {
+      "epoch": 0.008935335366445873,
+      "grad_norm": 1.1668083667755127,
+      "learning_rate": 0.0001,
+      "loss": 1.5261,
+      "step": 230
+    },
+    {
+      "epoch": 0.009323828208465258,
+      "grad_norm": 0.719367265701294,
+      "learning_rate": 0.0001,
+      "loss": 1.4151,
+      "step": 240
+    },
+    {
+      "epoch": 0.009712321050484645,
+      "grad_norm": 0.7443403005599976,
+      "learning_rate": 0.0001,
+      "loss": 1.3723,
+      "step": 250
+    },
+    {
+      "epoch": 0.010100813892504031,
+      "grad_norm": 0.8915978670120239,
+      "learning_rate": 0.0001,
+      "loss": 1.5165,
+      "step": 260
+    },
+    {
+      "epoch": 0.010489306734523416,
+      "grad_norm": 0.7369945049285889,
+      "learning_rate": 0.0001,
+      "loss": 1.4225,
+      "step": 270
+    },
+    {
+      "epoch": 0.010877799576542802,
+      "grad_norm": 1.2632057666778564,
+      "learning_rate": 0.0001,
+      "loss": 1.3462,
+      "step": 280
+    },
+    {
+      "epoch": 0.011266292418562187,
+      "grad_norm": 1.6178512573242188,
+      "learning_rate": 0.0001,
+      "loss": 1.4283,
+      "step": 290
+    },
+    {
+      "epoch": 0.011654785260581575,
+      "grad_norm": 2.717789649963379,
+      "learning_rate": 0.0001,
+      "loss": 1.4773,
+      "step": 300
+    },
+    {
+      "epoch": 0.011654785260581575,
+      "eval_loss": 1.4709599018096924,
+      "eval_runtime": 410.5517,
+      "eval_samples_per_second": 1.269,
+      "eval_steps_per_second": 1.269,
+      "step": 300
+    },
+    {
+      "epoch": 0.01204327810260096,
+      "grad_norm": 3.8710834980010986,
+      "learning_rate": 0.0001,
+      "loss": 1.4183,
+      "step": 310
+    },
+    {
+      "epoch": 0.012431770944620345,
+      "grad_norm": 1.1690031290054321,
+      "learning_rate": 0.0001,
+      "loss": 1.6159,
+      "step": 320
+    },
+    {
+      "epoch": 0.012820263786639731,
+      "grad_norm": 1.422135829925537,
+      "learning_rate": 0.0001,
+      "loss": 1.5327,
+      "step": 330
+    },
+    {
+      "epoch": 0.013208756628659116,
+      "grad_norm": 1.353925347328186,
+      "learning_rate": 0.0001,
+      "loss": 1.4782,
+      "step": 340
+    },
+    {
+      "epoch": 0.013597249470678504,
+      "grad_norm": 0.8083727359771729,
+      "learning_rate": 0.0001,
+      "loss": 1.3199,
+      "step": 350
+    },
+    {
+      "epoch": 0.013985742312697889,
+      "grad_norm": 0.6409865021705627,
+      "learning_rate": 0.0001,
+      "loss": 1.3829,
+      "step": 360
+    },
+    {
+      "epoch": 0.014374235154717275,
+      "grad_norm": 1.9057331085205078,
+      "learning_rate": 0.0001,
+      "loss": 1.3386,
+      "step": 370
+    },
+    {
+      "epoch": 0.014568481575726967,
+      "eval_loss": 1.4621069431304932,
+      "eval_runtime": 410.8963,
+      "eval_samples_per_second": 1.268,
+      "eval_steps_per_second": 1.268,
+      "step": 375
+    },
+    {
+      "epoch": 0.01476272799673666,
+      "grad_norm": 1.4260625839233398,
+      "learning_rate": 0.0001,
+      "loss": 1.4447,
+      "step": 380
+    },
+    {
+      "epoch": 0.015151220838756045,
+      "grad_norm": 2.0252511501312256,
+      "learning_rate": 0.0001,
+      "loss": 1.4396,
+      "step": 390
+    },
+    {
+      "epoch": 0.015539713680775431,
+      "grad_norm": 1.5493030548095703,
+      "learning_rate": 0.0001,
+      "loss": 1.5377,
+      "step": 400
+    },
+    {
+      "epoch": 0.015928206522794818,
+      "grad_norm": 1.5620871782302856,
+      "learning_rate": 0.0001,
+      "loss": 1.5368,
+      "step": 410
+    },
+    {
+      "epoch": 0.016316699364814204,
+      "grad_norm": 1.8342182636260986,
+      "learning_rate": 0.0001,
+      "loss": 1.4932,
+      "step": 420
+    },
+    {
+      "epoch": 0.01670519220683359,
+      "grad_norm": 0.8918685913085938,
+      "learning_rate": 0.0001,
+      "loss": 1.4847,
+      "step": 430
+    },
+    {
+      "epoch": 0.017093685048852975,
+      "grad_norm": 1.4548940658569336,
+      "learning_rate": 0.0001,
+      "loss": 1.5184,
+      "step": 440
+    },
+    {
+      "epoch": 0.01748217789087236,
+      "grad_norm": 1.4839730262756348,
+      "learning_rate": 0.0001,
+      "loss": 1.4276,
+      "step": 450
+    },
+    {
+      "epoch": 0.01748217789087236,
+      "eval_loss": 1.4603033065795898,
+      "eval_runtime": 408.9003,
+      "eval_samples_per_second": 1.274,
+      "eval_steps_per_second": 1.274,
+      "step": 450
+    },
+    {
+      "epoch": 0.017870670732891746,
+      "grad_norm": 0.6719891428947449,
+      "learning_rate": 0.0001,
+      "loss": 1.3042,
+      "step": 460
+    },
+    {
+      "epoch": 0.01825916357491113,
+      "grad_norm": 0.8530905246734619,
+      "learning_rate": 0.0001,
+      "loss": 1.454,
+      "step": 470
+    },
+    {
+      "epoch": 0.018647656416930516,
+      "grad_norm": 0.8087925910949707,
+      "learning_rate": 0.0001,
+      "loss": 1.4225,
+      "step": 480
+    },
+    {
+      "epoch": 0.019036149258949905,
+      "grad_norm": 2.091627359390259,
+      "learning_rate": 0.0001,
+      "loss": 1.4617,
+      "step": 490
+    },
+    {
+      "epoch": 0.01942464210096929,
+      "grad_norm": 2.1747212409973145,
+      "learning_rate": 0.0001,
+      "loss": 1.5124,
+      "step": 500
+    },
+    {
+      "epoch": 0.019813134942988676,
+      "grad_norm": 1.7147002220153809,
+      "learning_rate": 0.0001,
+      "loss": 1.4442,
+      "step": 510
+    },
+    {
+      "epoch": 0.020201627785008062,
+      "grad_norm": 0.7326516509056091,
+      "learning_rate": 0.0001,
+      "loss": 1.4376,
+      "step": 520
+    },
+    {
+      "epoch": 0.020395874206017753,
+      "eval_loss": 1.4605984687805176,
+      "eval_runtime": 418.9148,
+      "eval_samples_per_second": 1.244,
+      "eval_steps_per_second": 1.244,
+      "step": 525
+    },
+    {
+      "epoch": 0.020590120627027447,
+      "grad_norm": 1.7703779935836792,
+      "learning_rate": 0.0001,
+      "loss": 1.4911,
+      "step": 530
+    },
+    {
+      "epoch": 0.020978613469046833,
+      "grad_norm": 0.8552814722061157,
+      "learning_rate": 0.0001,
+      "loss": 1.4079,
+      "step": 540
+    },
+    {
+      "epoch": 0.021367106311066218,
+      "grad_norm": 1.0003011226654053,
+      "learning_rate": 0.0001,
+      "loss": 1.5769,
+      "step": 550
+    },
+    {
+      "epoch": 0.021755599153085604,
+      "grad_norm": 1.0176352262496948,
+      "learning_rate": 0.0001,
+      "loss": 1.2431,
+      "step": 560
+    },
+    {
+      "epoch": 0.02214409199510499,
+      "grad_norm": 1.8341186046600342,
+      "learning_rate": 0.0001,
+      "loss": 1.5053,
+      "step": 570
+    },
+    {
+      "epoch": 0.022532584837124375,
+      "grad_norm": 0.7317614555358887,
+      "learning_rate": 0.0001,
+      "loss": 1.5135,
+      "step": 580
+    },
+    {
+      "epoch": 0.02292107767914376,
+      "grad_norm": 1.3072996139526367,
+      "learning_rate": 0.0001,
+      "loss": 1.2935,
+      "step": 590
+    },
+    {
+      "epoch": 0.02330957052116315,
+      "grad_norm": 0.5384438633918762,
+      "learning_rate": 0.0001,
+      "loss": 1.4028,
+      "step": 600
+    },
+    {
+      "epoch": 0.02330957052116315,
+      "eval_loss": 1.457323670387268,
+      "eval_runtime": 419.1828,
+      "eval_samples_per_second": 1.243,
+      "eval_steps_per_second": 1.243,
+      "step": 600
+    },
+    {
+      "epoch": 0.023698063363182535,
+      "grad_norm": 0.6213059425354004,
+      "learning_rate": 0.0001,
+      "loss": 1.3461,
+      "step": 610
+    },
+    {
+      "epoch": 0.02408655620520192,
+      "grad_norm": 0.9022939801216125,
+      "learning_rate": 0.0001,
+      "loss": 1.458,
+      "step": 620
+    },
+    {
+      "epoch": 0.024475049047221305,
+      "grad_norm": 1.511841893196106,
+      "learning_rate": 0.0001,
+      "loss": 1.3387,
+      "step": 630
+    },
+    {
+      "epoch": 0.02486354188924069,
+      "grad_norm": 1.193332552909851,
+      "learning_rate": 0.0001,
+      "loss": 1.25,
+      "step": 640
+    },
+    {
+      "epoch": 0.025252034731260076,
+      "grad_norm": 0.664730429649353,
+      "learning_rate": 0.0001,
+      "loss": 1.4608,
+      "step": 650
+    },
+    {
+      "epoch": 0.025640527573279462,
+      "grad_norm": 0.9817675352096558,
+      "learning_rate": 0.0001,
+      "loss": 1.4694,
+      "step": 660
+    },
+    {
+      "epoch": 0.026029020415298847,
+      "grad_norm": 0.8713122606277466,
+      "learning_rate": 0.0001,
+      "loss": 1.6154,
+      "step": 670
+    },
+    {
+      "epoch": 0.026223266836308542,
+      "eval_loss": 1.4552098512649536,
+      "eval_runtime": 418.233,
+      "eval_samples_per_second": 1.246,
+      "eval_steps_per_second": 1.246,
+      "step": 675
+    },
+    {
+      "epoch": 0.026417513257318233,
+      "grad_norm": 0.8656709790229797,
+      "learning_rate": 0.0001,
+      "loss": 1.6028,
+      "step": 680
+    },
+    {
+      "epoch": 0.02680600609933762,
+      "grad_norm": 0.7827064990997314,
+      "learning_rate": 0.0001,
+      "loss": 1.454,
+      "step": 690
+    },
+    {
+      "epoch": 0.027194498941357007,
+      "grad_norm": 0.8780921101570129,
+      "learning_rate": 0.0001,
+      "loss": 1.377,
+      "step": 700
+    },
+    {
+      "epoch": 0.027582991783376393,
+      "grad_norm": 0.664682149887085,
+      "learning_rate": 0.0001,
+      "loss": 1.3761,
+      "step": 710
+    },
+    {
+      "epoch": 0.027971484625395778,
+      "grad_norm": 1.6883013248443604,
+      "learning_rate": 0.0001,
+      "loss": 1.293,
+      "step": 720
+    },
+    {
+      "epoch": 0.028359977467415164,
+      "grad_norm": 0.6659910082817078,
+      "learning_rate": 0.0001,
+      "loss": 1.3595,
+      "step": 730
+    },
+    {
+      "epoch": 0.02874847030943455,
+      "grad_norm": 1.0495606660842896,
+      "learning_rate": 0.0001,
+      "loss": 1.3881,
+      "step": 740
+    },
+    {
+      "epoch": 0.029136963151453935,
+      "grad_norm": 2.0675432682037354,
+      "learning_rate": 0.0001,
+      "loss": 1.3353,
+      "step": 750
+    },
+    {
+      "epoch": 0.029136963151453935,
+      "eval_loss": 1.4496526718139648,
+      "eval_runtime": 412.105,
+      "eval_samples_per_second": 1.264,
+      "eval_steps_per_second": 1.264,
+      "step": 750
+    },
+    {
+      "epoch": 0.02952545599347332,
+      "grad_norm": 2.147975444793701,
+      "learning_rate": 0.0001,
+      "loss": 1.4715,
+      "step": 760
+    },
+    {
+      "epoch": 0.029913948835492706,
+      "grad_norm": 1.4400185346603394,
+      "learning_rate": 0.0001,
+      "loss": 1.704,
+      "step": 770
+    },
+    {
+      "epoch": 0.03030244167751209,
+      "grad_norm": 0.5840633511543274,
+      "learning_rate": 0.0001,
+      "loss": 1.2531,
+      "step": 780
+    },
+    {
+      "epoch": 0.030690934519531476,
+      "grad_norm": 1.9958975315093994,
+      "learning_rate": 0.0001,
+      "loss": 1.6409,
+      "step": 790
+    },
+    {
+      "epoch": 0.031079427361550862,
+      "grad_norm": 0.4322706460952759,
+      "learning_rate": 0.0001,
+      "loss": 1.3866,
+      "step": 800
+    },
+    {
+      "epoch": 0.03146792020357025,
+      "grad_norm": 0.9608808755874634,
+      "learning_rate": 0.0001,
+      "loss": 1.2862,
+      "step": 810
+    },
+    {
+      "epoch": 0.031856413045589636,
+      "grad_norm": 1.0402257442474365,
+      "learning_rate": 0.0001,
+      "loss": 1.4305,
+      "step": 820
+    },
+    {
+      "epoch": 0.03205065946659933,
+      "eval_loss": 1.4473251104354858,
+      "eval_runtime": 410.7169,
+      "eval_samples_per_second": 1.269,
+      "eval_steps_per_second": 1.269,
+      "step": 825
+    },
+    {
+      "epoch": 0.03224490588760902,
+      "grad_norm": 0.6171532273292542,
+      "learning_rate": 0.0001,
+      "loss": 1.2107,
+      "step": 830
+    },
+    {
+      "epoch": 0.03263339872962841,
+      "grad_norm": 1.6381995677947998,
+      "learning_rate": 0.0001,
+      "loss": 1.5369,
+      "step": 840
+    },
+    {
+      "epoch": 0.03302189157164779,
+      "grad_norm": 0.5398985743522644,
+      "learning_rate": 0.0001,
+      "loss": 1.4939,
+      "step": 850
+    },
+    {
+      "epoch": 0.03341038441366718,
+      "grad_norm": 1.1927576065063477,
+      "learning_rate": 0.0001,
+      "loss": 1.459,
+      "step": 860
+    },
+    {
+      "epoch": 0.03379887725568657,
+      "grad_norm": 0.5355756878852844,
+      "learning_rate": 0.0001,
+      "loss": 1.4591,
+      "step": 870
+    },
+    {
+      "epoch": 0.03418737009770595,
+      "grad_norm": 1.0324468612670898,
+      "learning_rate": 0.0001,
+      "loss": 1.4079,
+      "step": 880
+    },
+    {
+      "epoch": 0.03457586293972534,
+      "grad_norm": 0.9082580804824829,
+      "learning_rate": 0.0001,
+      "loss": 1.4703,
+      "step": 890
+    },
+    {
+      "epoch": 0.03496435578174472,
+      "grad_norm": 1.0036635398864746,
+      "learning_rate": 0.0001,
+      "loss": 1.2948,
+      "step": 900
+    },
+    {
+      "epoch": 0.03496435578174472,
+      "eval_loss": 1.4607137441635132,
+      "eval_runtime": 418.5246,
+      "eval_samples_per_second": 1.245,
+      "eval_steps_per_second": 1.245,
+      "step": 900
+    },
+    {
+      "epoch": 0.03535284862376411,
+      "grad_norm": 0.7732622027397156,
+      "learning_rate": 0.0001,
+      "loss": 1.5017,
+      "step": 910
+    },
+    {
+      "epoch": 0.03574134146578349,
+      "grad_norm": 0.7425190806388855,
+      "learning_rate": 0.0001,
+      "loss": 1.4275,
+      "step": 920
+    },
+    {
+      "epoch": 0.03612983430780288,
+      "grad_norm": 0.6782093644142151,
+      "learning_rate": 0.0001,
+      "loss": 1.4424,
+      "step": 930
+    },
+    {
+      "epoch": 0.03651832714982226,
+      "grad_norm": 0.6914064288139343,
+      "learning_rate": 0.0001,
+      "loss": 1.49,
+      "step": 940
+    },
+    {
+      "epoch": 0.03690681999184165,
+      "grad_norm": 1.2722946405410767,
+      "learning_rate": 0.0001,
+      "loss": 1.4737,
+      "step": 950
+    },
+    {
+      "epoch": 0.03729531283386103,
+      "grad_norm": 0.9967614412307739,
+      "learning_rate": 0.0001,
+      "loss": 1.3731,
+      "step": 960
+    },
+    {
+      "epoch": 0.03768380567588042,
+      "grad_norm": 0.5614752173423767,
+      "learning_rate": 0.0001,
+      "loss": 1.3554,
+      "step": 970
+    },
+    {
+      "epoch": 0.037878052096890116,
+      "eval_loss": 1.4594156742095947,
+      "eval_runtime": 414.1638,
+      "eval_samples_per_second": 1.258,
+      "eval_steps_per_second": 1.258,
+      "step": 975
+    },
+    {
+      "epoch": 0.03807229851789981,
+      "grad_norm": 1.496825933456421,
+      "learning_rate": 0.0001,
+      "loss": 1.3043,
+      "step": 980
+    },
+    {
+      "epoch": 0.03846079135991919,
+      "grad_norm": 0.5324123501777649,
+      "learning_rate": 0.0001,
+      "loss": 1.5467,
+      "step": 990
+    },
+    {
+      "epoch": 0.03884928420193858,
+      "grad_norm": 2.828305959701538,
+      "learning_rate": 0.0001,
+      "loss": 1.4766,
+      "step": 1000
+    },
+    {
+      "epoch": 0.039237777043957964,
+      "grad_norm": 1.0788389444351196,
+      "learning_rate": 0.0001,
+      "loss": 1.6391,
+      "step": 1010
+    },
+    {
+      "epoch": 0.03962626988597735,
+      "grad_norm": 1.3913893699645996,
+      "learning_rate": 0.0001,
+      "loss": 1.36,
+      "step": 1020
+    },
+    {
+      "epoch": 0.040014762727996735,
+      "grad_norm": 1.0683279037475586,
+      "learning_rate": 0.0001,
+      "loss": 1.3993,
+      "step": 1030
+    },
+    {
+      "epoch": 0.040403255570016124,
+      "grad_norm": 0.14315283298492432,
+      "learning_rate": 0.0001,
+      "loss": 1.3594,
+      "step": 1040
+    },
+    {
+      "epoch": 0.040791748412035506,
+      "grad_norm": 1.996098518371582,
+      "learning_rate": 0.0001,
+      "loss": 1.3501,
+      "step": 1050
+    },
+    {
+      "epoch": 0.040791748412035506,
+      "eval_loss": 1.4564799070358276,
+      "eval_runtime": 407.4512,
+      "eval_samples_per_second": 1.279,
+      "eval_steps_per_second": 1.279,
+      "step": 1050
+    },
+    {
+      "epoch": 0.041180241254054895,
+      "grad_norm": 1.9238131046295166,
+      "learning_rate": 1e-05,
+      "loss": 1.3409,
+      "step": 1060
+    },
+    {
+      "epoch": 0.04156873409607428,
+      "grad_norm": 0.561337947845459,
+      "learning_rate": 1e-05,
+      "loss": 1.3931,
+      "step": 1070
+    },
+    {
+      "epoch": 0.041957226938093665,
+      "grad_norm": 0.6750600934028625,
+      "learning_rate": 1e-05,
+      "loss": 1.3055,
+      "step": 1080
+    },
+    {
+      "epoch": 0.042345719780113054,
+      "grad_norm": 1.6704535484313965,
+      "learning_rate": 1e-05,
+      "loss": 1.4371,
+      "step": 1090
+    },
+    {
+      "epoch": 0.042734212622132436,
+      "grad_norm": 0.6073994636535645,
+      "learning_rate": 1e-05,
+      "loss": 1.5461,
+      "step": 1100
+    },
+    {
+      "epoch": 0.043122705464151825,
+      "grad_norm": 1.1396293640136719,
+      "learning_rate": 1e-05,
+      "loss": 1.4749,
+      "step": 1110
+    },
+    {
+      "epoch": 0.04351119830617121,
+      "grad_norm": 0.6748817563056946,
+      "learning_rate": 1e-05,
+      "loss": 1.4803,
+      "step": 1120
+    },
+    {
+      "epoch": 0.0437054447271809,
+      "eval_loss": 1.4066863059997559,
+      "eval_runtime": 198.1965,
+      "eval_samples_per_second": 2.629,
+      "eval_steps_per_second": 2.629,
+      "step": 1125
+    },
+    {
+      "epoch": 0.043899691148190596,
+      "grad_norm": 0.8130941987037659,
+      "learning_rate": 1e-05,
+      "loss": 1.4579,
+      "step": 1130
+    },
+    {
+      "epoch": 0.04428818399020998,
+      "grad_norm": 0.5348241329193115,
+      "learning_rate": 1e-05,
+      "loss": 1.3988,
+      "step": 1140
+    },
+    {
+      "epoch": 0.04467667683222937,
+      "grad_norm": 0.6961309313774109,
+      "learning_rate": 1e-05,
+      "loss": 1.4051,
+      "step": 1150
+    },
+    {
+      "epoch": 0.04506516967424875,
+      "grad_norm": 0.8562794923782349,
+      "learning_rate": 1e-05,
+      "loss": 1.3631,
+      "step": 1160
+    },
+    {
+      "epoch": 0.04545366251626814,
+      "grad_norm": 0.6999790668487549,
+      "learning_rate": 1e-05,
+      "loss": 1.4532,
+      "step": 1170
+    },
+    {
+      "epoch": 0.04584215535828752,
+      "grad_norm": 0.5127655267715454,
+      "learning_rate": 1e-05,
+      "loss": 1.4715,
+      "step": 1180
+    },
+    {
+      "epoch": 0.04623064820030691,
+      "grad_norm": 1.5171382427215576,
+      "learning_rate": 1e-05,
+      "loss": 1.2901,
+      "step": 1190
+    },
+    {
+      "epoch": 0.0466191410423263,
+      "grad_norm": 0.7225420475006104,
+      "learning_rate": 1e-05,
+      "loss": 1.2778,
+      "step": 1200
+    },
+    {
+      "epoch": 0.0466191410423263,
+      "eval_loss": 1.4001317024230957,
+      "eval_runtime": 199.2632,
+      "eval_samples_per_second": 2.615,
+      "eval_steps_per_second": 2.615,
+      "step": 1200
+    },
+    {
+      "epoch": 0.04700763388434568,
+      "grad_norm": 1.5108428001403809,
+      "learning_rate": 1e-05,
+      "loss": 1.6175,
+      "step": 1210
+    },
+    {
+      "epoch": 0.04739612672636507,
+      "grad_norm": 1.1392805576324463,
+      "learning_rate": 1e-05,
+      "loss": 1.3466,
+      "step": 1220
+    },
+    {
+      "epoch": 0.04778461956838445,
+      "grad_norm": 0.94669109582901,
+      "learning_rate": 1e-05,
+      "loss": 1.3415,
+      "step": 1230
+    },
+    {
+      "epoch": 0.04817311241040384,
+      "grad_norm": 0.8593105673789978,
+      "learning_rate": 1e-05,
+      "loss": 1.3334,
+      "step": 1240
+    },
+    {
+      "epoch": 0.04856160525242322,
+      "grad_norm": 0.8188263773918152,
+      "learning_rate": 1e-05,
+      "loss": 1.4602,
+      "step": 1250
+    },
+    {
+      "epoch": 0.04895009809444261,
+      "grad_norm": 0.6875782608985901,
+      "learning_rate": 1e-05,
+      "loss": 1.3261,
+      "step": 1260
+    },
+    {
+      "epoch": 0.04933859093646199,
+      "grad_norm": 1.8237006664276123,
+      "learning_rate": 1e-05,
+      "loss": 1.5862,
+      "step": 1270
+    },
+    {
+      "epoch": 0.04953283735747169,
+      "eval_loss": 1.3970199823379517,
+      "eval_runtime": 205.8088,
+      "eval_samples_per_second": 2.531,
+      "eval_steps_per_second": 2.531,
+      "step": 1275
+    },
+    {
+      "epoch": 0.04972708377848138,
+      "grad_norm": 1.319785237312317,
+      "learning_rate": 1e-05,
+      "loss": 1.3576,
+      "step": 1280
+    },
+    {
+      "epoch": 0.050115576620500764,
+      "grad_norm": 1.727789282798767,
+      "learning_rate": 1e-05,
+      "loss": 1.5409,
+      "step": 1290
+    },
+    {
+      "epoch": 0.05050406946252015,
+      "grad_norm": 0.9914244413375854,
+      "learning_rate": 1e-05,
+      "loss": 1.3503,
+      "step": 1300
+    },
+    {
+      "epoch": 0.05089256230453954,
+      "grad_norm": 1.8328955173492432,
+      "learning_rate": 1e-05,
+      "loss": 1.5384,
+      "step": 1310
+    },
+    {
+      "epoch": 0.051281055146558924,
+      "grad_norm": 1.7998759746551514,
+      "learning_rate": 1e-05,
+      "loss": 1.4807,
+      "step": 1320
+    },
+    {
+      "epoch": 0.05166954798857831,
+      "grad_norm": 1.53579843044281,
+      "learning_rate": 1e-05,
+      "loss": 1.4255,
+      "step": 1330
+    },
+    {
+      "epoch": 0.052058040830597695,
+      "grad_norm": 0.9572857022285461,
+      "learning_rate": 1e-05,
+      "loss": 1.4547,
+      "step": 1340
+    },
+    {
+      "epoch": 0.052446533672617084,
+      "grad_norm": 0.6299539804458618,
+      "learning_rate": 1e-05,
+      "loss": 1.2758,
+      "step": 1350
+    },
+    {
+      "epoch": 0.052446533672617084,
+      "eval_loss": 1.394976258277893,
+      "eval_runtime": 206.0718,
+      "eval_samples_per_second": 2.528,
+      "eval_steps_per_second": 2.528,
+      "step": 1350
+    },
+    {
+      "epoch": 0.052835026514636466,
+      "grad_norm": 1.1869505643844604,
+      "learning_rate": 1e-05,
+      "loss": 1.2709,
+      "step": 1360
+    },
+    {
+      "epoch": 0.053223519356655855,
+      "grad_norm": 0.5684358477592468,
+      "learning_rate": 1e-05,
+      "loss": 1.3306,
+      "step": 1370
+    },
+    {
+      "epoch": 0.05361201219867524,
+      "grad_norm": 0.5880847573280334,
+      "learning_rate": 1e-05,
+      "loss": 1.3093,
+      "step": 1380
+    },
+    {
+      "epoch": 0.054000505040694625,
+      "grad_norm": 0.6990231275558472,
+      "learning_rate": 1e-05,
+      "loss": 1.4534,
+      "step": 1390
+    },
+    {
+      "epoch": 0.054388997882714014,
+      "grad_norm": 1.0700093507766724,
+      "learning_rate": 1e-05,
+      "loss": 1.3294,
+      "step": 1400
+    },
+    {
+      "epoch": 0.054777490724733396,
+      "grad_norm": 1.044433832168579,
+      "learning_rate": 1e-05,
+      "loss": 1.4177,
+      "step": 1410
+    },
+    {
+      "epoch": 0.055165983566752785,
+      "grad_norm": 2.6891329288482666,
+      "learning_rate": 1e-05,
+      "loss": 1.4451,
+      "step": 1420
+    },
+    {
+      "epoch": 0.05536022998776247,
+      "eval_loss": 1.3934379816055298,
+      "eval_runtime": 204.0926,
+      "eval_samples_per_second": 2.553,
+      "eval_steps_per_second": 2.553,
+      "step": 1425
+    },
+    {
+      "epoch": 0.05555447640877217,
+      "grad_norm": 0.4769861698150635,
+      "learning_rate": 1e-05,
+      "loss": 1.3179,
+      "step": 1430
+    },
+    {
+      "epoch": 0.055942969250791556,
+      "grad_norm": 1.0731093883514404,
+      "learning_rate": 1e-05,
+      "loss": 1.47,
+      "step": 1440
+    },
+    {
+      "epoch": 0.05633146209281094,
+      "grad_norm": 1.016760230064392,
+      "learning_rate": 1e-05,
+      "loss": 1.5151,
+      "step": 1450
+    },
+    {
+      "epoch": 0.05671995493483033,
+      "grad_norm": 1.5259450674057007,
+      "learning_rate": 1e-05,
+      "loss": 1.4038,
+      "step": 1460
+    },
+    {
+      "epoch": 0.05710844777684971,
+      "grad_norm": 0.654501736164093,
+      "learning_rate": 1e-05,
+      "loss": 1.3135,
+      "step": 1470
+    },
+    {
+      "epoch": 0.0574969406188691,
+      "grad_norm": 0.6827269196510315,
+      "learning_rate": 1e-05,
+      "loss": 1.2978,
+      "step": 1480
+    },
+    {
+      "epoch": 0.05788543346088848,
+      "grad_norm": 0.5111151933670044,
+      "learning_rate": 1e-05,
+      "loss": 1.4352,
+      "step": 1490
+    },
+    {
+      "epoch": 0.05827392630290787,
+      "grad_norm": 1.9571446180343628,
+      "learning_rate": 1e-05,
+      "loss": 1.4764,
+      "step": 1500
+    },
+    {
+      "epoch": 0.05827392630290787,
+      "eval_loss": 1.3912627696990967,
+      "eval_runtime": 204.5081,
+      "eval_samples_per_second": 2.548,
+      "eval_steps_per_second": 2.548,
+      "step": 1500
+    },
+    {
+      "epoch": 0.05866241914492726,
+      "grad_norm": 0.8712412714958191,
+      "learning_rate": 1e-05,
+      "loss": 1.3778,
+      "step": 1510
+    },
+    {
+      "epoch": 0.05905091198694664,
+      "grad_norm": 0.7130087018013,
+      "learning_rate": 1e-05,
+      "loss": 1.3266,
+      "step": 1520
+    },
+    {
+      "epoch": 0.05943940482896603,
+      "grad_norm": 1.6288388967514038,
+      "learning_rate": 1e-05,
+      "loss": 1.4783,
+      "step": 1530
+    },
+    {
+      "epoch": 0.05982789767098541,
+      "grad_norm": 2.629760503768921,
+      "learning_rate": 1e-05,
+      "loss": 1.6038,
+      "step": 1540
+    },
+    {
+      "epoch": 0.0602163905130048,
+      "grad_norm": 1.0394636392593384,
+      "learning_rate": 1e-05,
+      "loss": 1.4683,
+      "step": 1550
+    },
+    {
+      "epoch": 0.06060488335502418,
+      "grad_norm": 1.128451943397522,
+      "learning_rate": 1e-05,
+      "loss": 1.4578,
+      "step": 1560
+    },
+    {
+      "epoch": 0.06099337619704357,
+      "grad_norm": 2.473900079727173,
+      "learning_rate": 1e-05,
+      "loss": 1.4326,
+      "step": 1570
+    },
+    {
+      "epoch": 0.061187622618053265,
+      "eval_loss": 1.3889408111572266,
+      "eval_runtime": 205.5592,
+      "eval_samples_per_second": 2.535,
+      "eval_steps_per_second": 2.535,
+      "step": 1575
+    },
+    {
+      "epoch": 0.06138186903906295,
+      "grad_norm": 1.940373182296753,
+      "learning_rate": 1e-05,
+      "loss": 1.5117,
+      "step": 1580
+    },
+    {
+      "epoch": 0.06177036188108234,
+      "grad_norm": 0.7575955986976624,
+      "learning_rate": 1e-05,
+      "loss": 1.3894,
+      "step": 1590
+    },
+    {
+      "epoch": 0.062158854723101724,
+      "grad_norm": 1.4801169633865356,
+      "learning_rate": 1e-05,
+      "loss": 1.3132,
+      "step": 1600
+    },
+    {
+      "epoch": 0.0625473475651211,
+      "grad_norm": 1.291632890701294,
+      "learning_rate": 1e-05,
+      "loss": 1.3286,
+      "step": 1610
+    },
+    {
+      "epoch": 0.0629358404071405,
+      "grad_norm": 1.9607435464859009,
+      "learning_rate": 1e-05,
+      "loss": 1.3005,
+      "step": 1620
+    },
+    {
+      "epoch": 0.06332433324915988,
+      "grad_norm": 0.8362483382225037,
+      "learning_rate": 1e-05,
+      "loss": 1.4172,
+      "step": 1630
+    },
+    {
+      "epoch": 0.06371282609117927,
+      "grad_norm": 1.3649120330810547,
+      "learning_rate": 1e-05,
+      "loss": 1.6757,
+      "step": 1640
+    },
+    {
+      "epoch": 0.06410131893319866,
+      "grad_norm": 1.0758274793624878,
+      "learning_rate": 1e-05,
+      "loss": 1.3867,
+      "step": 1650
+    },
+    {
+      "epoch": 0.06410131893319866,
+      "eval_loss": 1.3888965845108032,
+      "eval_runtime": 205.6882,
+      "eval_samples_per_second": 2.533,
+      "eval_steps_per_second": 2.533,
+      "step": 1650
+    },
+    {
+      "epoch": 0.06448981177521804,
+      "grad_norm": 0.8754805326461792,
+      "learning_rate": 1e-05,
+      "loss": 1.3389,
+      "step": 1660
+    },
+    {
+      "epoch": 0.06487830461723743,
+      "grad_norm": 0.7831467986106873,
+      "learning_rate": 1e-05,
+      "loss": 1.4257,
+      "step": 1670
+    },
+    {
+      "epoch": 0.06526679745925681,
+      "grad_norm": 0.4581933915615082,
+      "learning_rate": 1e-05,
+      "loss": 1.3556,
+      "step": 1680
+    },
+    {
+      "epoch": 0.0656552903012762,
+      "grad_norm": 0.9837825894355774,
+      "learning_rate": 1e-05,
+      "loss": 1.3184,
+      "step": 1690
+    },
+    {
+      "epoch": 0.06604378314329558,
+      "grad_norm": 1.005288004875183,
+      "learning_rate": 1e-05,
+      "loss": 1.2944,
+      "step": 1700
+    },
+    {
+      "epoch": 0.06643227598531497,
+      "grad_norm": 0.9397820234298706,
+      "learning_rate": 1e-05,
+      "loss": 1.4305,
+      "step": 1710
+    },
+    {
+      "epoch": 0.06682076882733436,
+      "grad_norm": 2.7833900451660156,
+      "learning_rate": 1e-05,
+      "loss": 1.3273,
+      "step": 1720
+    },
+    {
+      "epoch": 0.06701501524834405,
+      "eval_loss": 1.3884316682815552,
+      "eval_runtime": 206.1573,
+      "eval_samples_per_second": 2.527,
+      "eval_steps_per_second": 2.527,
+      "step": 1725
+    },
+    {
+      "epoch": 0.06720926166935375,
+      "grad_norm": 1.1208202838897705,
+      "learning_rate": 1e-05,
+      "loss": 1.2229,
+      "step": 1730
+    },
+    {
+      "epoch": 0.06759775451137313,
+      "grad_norm": 0.5742992758750916,
+      "learning_rate": 1e-05,
+      "loss": 1.3349,
+      "step": 1740
+    },
+    {
+      "epoch": 0.06798624735339251,
+      "grad_norm": 0.7946904897689819,
+      "learning_rate": 1e-05,
+      "loss": 1.3682,
+      "step": 1750
+    },
+    {
+      "epoch": 0.0683747401954119,
+      "grad_norm": 0.7263549566268921,
+      "learning_rate": 1e-05,
+      "loss": 1.5025,
+      "step": 1760
+    },
+    {
+      "epoch": 0.06876323303743129,
+      "grad_norm": 0.8954797387123108,
+      "learning_rate": 1e-05,
+      "loss": 1.4383,
+      "step": 1770
+    },
+    {
+      "epoch": 0.06915172587945068,
+      "grad_norm": 0.6124446392059326,
+      "learning_rate": 1e-05,
+      "loss": 1.3322,
+      "step": 1780
+    },
+    {
+      "epoch": 0.06954021872147005,
+      "grad_norm": 1.140678882598877,
+      "learning_rate": 1e-05,
+      "loss": 1.5233,
+      "step": 1790
+    },
+    {
+      "epoch": 0.06992871156348944,
+      "grad_norm": 7.1586689949035645,
+      "learning_rate": 1e-05,
+      "loss": 1.3691,
+      "step": 1800
+    },
+    {
+      "epoch": 0.06992871156348944,
+      "eval_loss": 1.387437105178833,
+      "eval_runtime": 204.4312,
+      "eval_samples_per_second": 2.549,
+      "eval_steps_per_second": 2.549,
+      "step": 1800
+    },
+    {
+      "epoch": 0.07031720440550883,
+      "grad_norm": 0.634140133857727,
+      "learning_rate": 1e-05,
+      "loss": 1.3735,
+      "step": 1810
+    },
+    {
+      "epoch": 0.07070569724752822,
+      "grad_norm": 0.7632227540016174,
+      "learning_rate": 1e-05,
+      "loss": 1.3542,
+      "step": 1820
+    },
+    {
+      "epoch": 0.07109419008954761,
+      "grad_norm": 0.7211370468139648,
+      "learning_rate": 1e-05,
+      "loss": 1.2832,
+      "step": 1830
+    },
+    {
+      "epoch": 0.07148268293156698,
+      "grad_norm": 0.7608075737953186,
+      "learning_rate": 1e-05,
+      "loss": 1.5292,
+      "step": 1840
+    },
+    {
+      "epoch": 0.07187117577358637,
+      "grad_norm": 0.8131744265556335,
+      "learning_rate": 1e-05,
+      "loss": 1.4005,
+      "step": 1850
+    },
+    {
+      "epoch": 0.07225966861560576,
+      "grad_norm": 0.6415278911590576,
+      "learning_rate": 1e-05,
+      "loss": 1.4455,
+      "step": 1860
+    },
+    {
+      "epoch": 0.07264816145762515,
+      "grad_norm": 2.333056688308716,
+      "learning_rate": 1e-05,
+      "loss": 1.4367,
+      "step": 1870
+    },
+    {
+      "epoch": 0.07284240787863483,
+      "eval_loss": 1.3867840766906738,
+      "eval_runtime": 205.2163,
+      "eval_samples_per_second": 2.539,
+      "eval_steps_per_second": 2.539,
+      "step": 1875
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 3600,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 75,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 8.331378819072e+16,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1357f2f95116f571e7f20f337ece9446f8849c2418f238a6195a193e9e17bd5a
+size 7608

zero_to_fp32.py

@@ -0,0 +1,674 @@
+#!/usr/bin/env python
+
+# Copyright (c) Microsoft Corporation.
+# SPDX-License-Identifier: Apache-2.0
+
+# DeepSpeed Team
+
+# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
+# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
+# the future. Once extracted, the weights don't require DeepSpeed and can be used in any
+# application.
+#
+# example:
+#   python zero_to_fp32.py . output_dir/
+#   or
+#   python zero_to_fp32.py . output_dir/ --safe_serialization
+
+import argparse
+import torch
+import glob
+import math
+import os
+import re
+import json
+from tqdm import tqdm
+from collections import OrderedDict
+from dataclasses import dataclass
+
+# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
+# DeepSpeed data structures it has to be available in the current python environment.
+from deepspeed.utils import logger
+from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
+                                            FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
+                                            FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
+
+
+@dataclass
+class zero_model_state:
+    buffers: dict()
+    param_shapes: dict()
+    shared_params: list
+    ds_version: int
+    frozen_param_shapes: dict()
+    frozen_param_fragments: dict()
+
+
+debug = 0
+
+# load to cpu
+device = torch.device('cpu')
+
+
+def atoi(text):
+    return int(text) if text.isdigit() else text
+
+
+def natural_keys(text):
+    '''
+    alist.sort(key=natural_keys) sorts in human order
+    http://nedbatchelder.com/blog/200712/human_sorting.html
+    (See Toothy's implementation in the comments)
+    '''
+    return [atoi(c) for c in re.split(r'(\d+)', text)]
+
+
+def get_model_state_file(checkpoint_dir, zero_stage):
+    if not os.path.isdir(checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
+
+    # there should be only one file
+    if zero_stage <= 2:
+        file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
+    elif zero_stage == 3:
+        file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
+
+    if not os.path.exists(file):
+        raise FileNotFoundError(f"can't find model states file at '{file}'")
+
+    return file
+
+
+def get_checkpoint_files(checkpoint_dir, glob_pattern):
+    # XXX: need to test that this simple glob rule works for multi-node setup too
+    ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
+
+    if len(ckpt_files) == 0:
+        raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
+
+    return ckpt_files
+
+
+def get_optim_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
+
+
+def get_model_state_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
+
+
+def parse_model_states(files):
+    zero_model_states = []
+    for file in files:
+        state_dict = torch.load(file, map_location=device)
+
+        if BUFFER_NAMES not in state_dict:
+            raise ValueError(f"{file} is not a model state checkpoint")
+        buffer_names = state_dict[BUFFER_NAMES]
+        if debug:
+            print("Found buffers:", buffer_names)
+
+        # recover just the buffers while restoring them to fp32 if they were saved in fp16
+        buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
+        param_shapes = state_dict[PARAM_SHAPES]
+
+        # collect parameters that are included in param_shapes
+        param_names = []
+        for s in param_shapes:
+            for name in s.keys():
+                param_names.append(name)
+
+        # update with frozen parameters
+        frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
+        if frozen_param_shapes is not None:
+            if debug:
+                print(f"Found frozen_param_shapes: {frozen_param_shapes}")
+            param_names += list(frozen_param_shapes.keys())
+
+        # handle shared params
+        shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
+
+        ds_version = state_dict.get(DS_VERSION, None)
+
+        frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
+
+        z_model_state = zero_model_state(buffers=buffers,
+                                         param_shapes=param_shapes,
+                                         shared_params=shared_params,
+                                         ds_version=ds_version,
+                                         frozen_param_shapes=frozen_param_shapes,
+                                         frozen_param_fragments=frozen_param_fragments)
+        zero_model_states.append(z_model_state)
+
+    return zero_model_states
+
+
+def parse_optim_states(files, ds_checkpoint_dir):
+    total_files = len(files)
+    state_dicts = []
+    for f in files:
+        state_dict = torch.load(f, map_location=device)
+        # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
+        # and also handle the case where it was already removed by another helper script
+        state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
+        state_dicts.append(state_dict)
+
+    if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
+        raise ValueError(f"{files[0]} is not a zero checkpoint")
+    zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
+    world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
+
+    # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
+    # parameters can be different from data parallelism for non-expert parameters. So we can just
+    # use the max of the partition_count to get the dp world_size.
+
+    if type(world_size) is list:
+        world_size = max(world_size)
+
+    if world_size != total_files:
+        raise ValueError(
+            f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
+            "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
+        )
+
+    # the groups are named differently in each stage
+    if zero_stage <= 2:
+        fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
+    elif zero_stage == 3:
+        fp32_groups_key = FP32_FLAT_GROUPS
+    else:
+        raise ValueError(f"unknown zero stage {zero_stage}")
+
+    if zero_stage <= 2:
+        fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
+    elif zero_stage == 3:
+        # if there is more than one param group, there will be multiple flattened tensors - one
+        # flattened tensor per group - for simplicity merge them into a single tensor
+        #
+        # XXX: could make the script more memory efficient for when there are multiple groups - it
+        # will require matching the sub-lists of param_shapes for each param group flattened tensor
+
+        fp32_flat_groups = [
+            torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
+        ]
+
+    return zero_stage, world_size, fp32_flat_groups
+
+
+def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters):
+    """
+    Returns fp32 state_dict reconstructed from ds checkpoint
+
+    Args:
+        - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
+
+    """
+    print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
+
+    optim_files = get_optim_files(ds_checkpoint_dir)
+    zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
+    print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
+
+    model_files = get_model_state_files(ds_checkpoint_dir)
+
+    zero_model_states = parse_model_states(model_files)
+    print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
+
+    if zero_stage <= 2:
+        return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+    elif zero_stage == 3:
+        return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+
+
+def _zero2_merge_frozen_params(state_dict, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+    frozen_param_fragments = zero_model_states[0].frozen_param_fragments
+
+    if debug:
+        num_elem = sum(s.numel() for s in frozen_param_shapes.values())
+        print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        state_dict[name] = frozen_param_fragments[name]
+
+        if debug:
+            print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _has_callable(obj, fn):
+    attr = getattr(obj, fn, None)
+    return callable(attr)
+
+
+def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+
+    # Reconstruction protocol:
+    #
+    # XXX: document this
+
+    if debug:
+        for i in range(world_size):
+            for j in range(len(fp32_flat_groups[0])):
+                print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
+
+    # XXX: memory usage doubles here (zero2)
+    num_param_groups = len(fp32_flat_groups[0])
+    merged_single_partition_of_fp32_groups = []
+    for i in range(num_param_groups):
+        merged_partitions = [sd[i] for sd in fp32_flat_groups]
+        full_single_fp32_vector = torch.cat(merged_partitions, 0)
+        merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
+    avail_numel = sum(
+        [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
+
+    if debug:
+        wanted_params = sum([len(shapes) for shapes in param_shapes])
+        wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
+        # not asserting if there is a mismatch due to possible padding
+        print(f"Have {avail_numel} numels to process.")
+        print(f"Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    total_numel = 0
+    total_params = 0
+    for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
+        offset = 0
+        avail_numel = full_single_fp32_vector.numel()
+        for name, shape in shapes.items():
+
+            unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
+            total_numel += unpartitioned_numel
+            total_params += 1
+
+            if debug:
+                print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+            state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
+            offset += unpartitioned_numel
+
+        # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
+        # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
+        # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
+        # live optimizer object, so we are checking that the numbers are within the right range
+        align_to = 2 * world_size
+
+        def zero2_align(x):
+            return align_to * math.ceil(x / align_to)
+
+        if debug:
+            print(f"original offset={offset}, avail_numel={avail_numel}")
+
+        offset = zero2_align(offset)
+        avail_numel = zero2_align(avail_numel)
+
+        if debug:
+            print(f"aligned  offset={offset}, avail_numel={avail_numel}")
+
+        # Sanity check
+        if offset != avail_numel:
+            raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _zero2_merge_frozen_params(state_dict, zero_model_states)
+
+    _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def zero3_partitioned_param_info(unpartitioned_numel, world_size):
+    remainder = unpartitioned_numel % world_size
+    padding_numel = (world_size - remainder) if remainder else 0
+    partitioned_numel = math.ceil(unpartitioned_numel / world_size)
+    return partitioned_numel, padding_numel
+
+
+def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    if debug:
+        for i in range(world_size):
+            num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
+            print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in zero_model_states[0].frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
+        state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
+
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+    avail_numel = fp32_flat_groups[0].numel() * world_size
+    # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
+    # param, re-consolidating each param, while dealing with padding if any
+
+    # merge list of dicts, preserving order
+    param_shapes = {k: v for d in param_shapes for k, v in d.items()}
+
+    if debug:
+        for i in range(world_size):
+            print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
+
+        wanted_params = len(param_shapes)
+        wanted_numel = sum(shape.numel() for shape in param_shapes.values())
+        # not asserting if there is a mismatch due to possible padding
+        avail_numel = fp32_flat_groups[0].numel() * world_size
+        print(f"Trainable params: Have {avail_numel} numels to process.")
+        print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    offset = 0
+    total_numel = 0
+    total_params = 0
+    for name, shape in tqdm(param_shapes.items(), desc='Gathering Sharded Weights'):
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+        total_params += 1
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+        # XXX: memory usage doubles here
+        state_dict[name] = torch.cat(
+            tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
+            0).narrow(0, 0, unpartitioned_numel).view(shape)
+        offset += partitioned_numel
+
+    offset *= world_size
+
+    # Sanity check
+    if offset != avail_numel:
+        raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
+
+    _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
+    ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
+    via a model hub.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder
+        - ``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``
+        - ``exclude_frozen_parameters``: exclude frozen parameters
+
+    Returns:
+        - pytorch ``state_dict``
+
+    Note: this approach may not work if your application doesn't have sufficient free CPU memory and
+    you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
+    the checkpoint.
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+        # do the training and checkpoint saving
+        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
+        model = model.cpu() # move to cpu
+        model.load_state_dict(state_dict)
+        # submit to model hub or save the model to share with others
+
+    In this example the ``model`` will no longer be usable in the deepspeed context of the same
+    application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
+
+    """
+    if tag is None:
+        latest_path = os.path.join(checkpoint_dir, 'latest')
+        if os.path.isfile(latest_path):
+            with open(latest_path, 'r') as fd:
+                tag = fd.read().strip()
+        else:
+            raise ValueError(f"Unable to find 'latest' file at {latest_path}")
+
+    ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
+
+    if not os.path.isdir(ds_checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
+
+    return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
+
+
+def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir,
+                                               output_dir,
+                                               max_shard_size="5GB",
+                                               safe_serialization=False,
+                                               tag=None,
+                                               exclude_frozen_parameters=False):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
+    loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``output_dir``: directory to the pytorch fp32 state_dict output files
+        - ``max_shard_size``: the maximum size for a checkpoint before being sharded, default value is 5GB
+        - ``safe_serialization``:  whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).
+        - ``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``
+        - ``exclude_frozen_parameters``: exclude frozen parameters
+    """
+    # Dependency pre-check
+    if safe_serialization:
+        try:
+            from safetensors.torch import save_file
+        except ImportError:
+            print('If you want to use `safe_serialization`, please `pip install safetensors`')
+            raise
+    if max_shard_size is not None:
+        try:
+            from huggingface_hub import split_torch_state_dict_into_shards
+        except ImportError:
+            print('If you want to use `max_shard_size`, please `pip install huggingface_hub`')
+            raise
+
+    # Convert zero checkpoint to state_dict
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters)
+
+    # Shard the model if it is too big.
+    weights_name = "model.safetensors" if safe_serialization else "pytorch_model.bin"
+    if max_shard_size is not None:
+        filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(".safetensors", "{suffix}.safetensors")
+        state_dict_split = split_torch_state_dict_into_shards(state_dict,
+                                                              filename_pattern=filename_pattern,
+                                                              max_shard_size=max_shard_size)
+    else:
+        from collections import namedtuple
+        StateDictSplit = namedtuple("StateDictSplit", ["is_sharded", "filename_to_tensors"])
+        state_dict_split = StateDictSplit(is_sharded=False,
+                                          filename_to_tensors={weights_name: list(state_dict.keys())})
+
+    # Save the model
+    filename_to_tensors = state_dict_split.filename_to_tensors.items()
+    for shard_file, tensors in tqdm(filename_to_tensors, desc="Saving checkpoint shards"):
+        shard = {tensor: state_dict[tensor].contiguous() for tensor in tensors}
+        output_path = os.path.join(output_dir, shard_file)
+        if safe_serialization:
+            save_file(shard, output_path, metadata={"format": "pt"})
+        else:
+            torch.save(shard, output_path)
+
+    # Save index if sharded
+    if state_dict_split.is_sharded:
+        index = {
+            "metadata": state_dict_split.metadata,
+            "weight_map": state_dict_split.tensor_to_filename,
+        }
+        save_index_file = "model.safetensors.index.json" if safe_serialization else "pytorch_model.bin.index.json"
+        save_index_file = os.path.join(output_dir, save_index_file)
+        with open(save_index_file, "w", encoding="utf-8") as f:
+            content = json.dumps(index, indent=2, sort_keys=True) + "\n"
+            f.write(content)
+
+
+def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
+    """
+    1. Put the provided model to cpu
+    2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
+    3. Load it into the provided model
+
+    Args:
+        - ``model``: the model object to update
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``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``
+
+    Returns:
+        - ``model`: modified model
+
+    Make sure you have plenty of CPU memory available before you call this function. If you don't
+    have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
+    conveniently placed for you in the checkpoint folder.
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
+        model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
+        # submit to model hub or save the model to share with others
+
+    Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
+    of the same application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    """
+    logger.info(f"Extracting fp32 weights")
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
+
+    logger.info(f"Overwriting model with fp32 weights")
+    model = model.cpu()
+    model.load_state_dict(state_dict, strict=False)
+
+    return model
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("checkpoint_dir",
+                        type=str,
+                        help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
+    parser.add_argument("output_dir",
+                        type=str,
+                        help="directory to the pytorch fp32 state_dict output files"
+                        "(e.g. path/checkpoint-12-output/)")
+    parser.add_argument(
+        "--max_shard_size",
+        type=str,
+        default="5GB",
+        help="The maximum size for a checkpoint before being sharded. Checkpoints shard will then be each of size"
+        "lower than this size. If expressed as a string, needs to be digits followed by a unit (like `5MB`"
+        "We default it to 5GB in order for models to be able to run easily on free-tier google colab instances"
+        "without CPU OOM issues.")
+    parser.add_argument(
+        "--safe_serialization",
+        default=False,
+        action='store_true',
+        help="Whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).")
+    parser.add_argument("-t",
+                        "--tag",
+                        type=str,
+                        default=None,
+                        help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
+    parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters")
+    parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
+    args = parser.parse_args()
+
+    debug = args.debug
+
+    convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir,
+                                               args.output_dir,
+                                               max_shard_size=args.max_shard_size,
+                                               safe_serialization=args.safe_serialization,
+                                               tag=args.tag,
+                                               exclude_frozen_parameters=args.exclude_frozen_parameters)