Delta-Vector
commited on
Upload Gemma-Nemo-C2-Chatml.yml
Browse files- Gemma-Nemo-C2-Chatml.yml +100 -0
Gemma-Nemo-C2-Chatml.yml
ADDED
@@ -0,0 +1,100 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
base_model: SillyTilly/google-gemma-2-9b
|
2 |
+
model_type: AutoModelForCausalLM
|
3 |
+
tokenizer_type: AutoTokenizer
|
4 |
+
|
5 |
+
plugins:
|
6 |
+
- axolotl.integrations.liger.LigerPlugin
|
7 |
+
liger_rope: true
|
8 |
+
liger_rms_norm: true
|
9 |
+
liger_swiglu: true
|
10 |
+
#liger_cross_entropy: true
|
11 |
+
liger_fused_linear_cross_entropy: true
|
12 |
+
|
13 |
+
load_in_8bit: false
|
14 |
+
load_in_4bit: false
|
15 |
+
strict: false
|
16 |
+
|
17 |
+
datasets:
|
18 |
+
- path: anthracite-core/c2_logs_16k_llama_v1.1
|
19 |
+
type: sharegpt
|
20 |
+
conversation: chatml
|
21 |
+
- path: anthracite-org/kalo-opus-instruct-22k-no-refusal
|
22 |
+
type: sharegpt
|
23 |
+
conversation: chatml
|
24 |
+
- path: Epiculous/SynthRP-Gens-v1.1-Filtered-n-Cleaned
|
25 |
+
type: sharegpt
|
26 |
+
conversation: chatml
|
27 |
+
- path: lodrick-the-lafted/kalo-opus-instruct-3k-filtered
|
28 |
+
type: sharegpt
|
29 |
+
conversation: chatml
|
30 |
+
- path: anthracite-org/nopm_claude_writing_fixed
|
31 |
+
type: sharegpt
|
32 |
+
conversation: chatml
|
33 |
+
- path: Epiculous/Synthstruct-Gens-v1.1-Filtered-n-Cleaned
|
34 |
+
type: sharegpt
|
35 |
+
conversation: chatml
|
36 |
+
- path: anthracite-org/kalo_opus_misc_240827
|
37 |
+
type: sharegpt
|
38 |
+
conversation: chatml
|
39 |
+
- path: anthracite-org/kalo_misc_part2
|
40 |
+
type: sharegpt
|
41 |
+
conversation: chatml
|
42 |
+
chat_template: chatml
|
43 |
+
shuffle_merged_datasets: false
|
44 |
+
default_system_message: "You are a helpful assistant that responds to the user."
|
45 |
+
dataset_prepared_path: /workspace/data/9b-fft-data
|
46 |
+
val_set_size: 0.0
|
47 |
+
output_dir: /workspace/data/9b-fft-out
|
48 |
+
|
49 |
+
sequence_len: 8192
|
50 |
+
sample_packing: true
|
51 |
+
eval_sample_packing: false
|
52 |
+
pad_to_sequence_len: true
|
53 |
+
|
54 |
+
adapter:
|
55 |
+
lora_model_dir:
|
56 |
+
lora_r:
|
57 |
+
lora_alpha:
|
58 |
+
lora_dropout:
|
59 |
+
lora_target_linear:
|
60 |
+
lora_fan_in_fan_out:
|
61 |
+
|
62 |
+
wandb_project: 9b-Nemo-config-fft
|
63 |
+
wandb_entity:
|
64 |
+
wandb_watch:
|
65 |
+
wandb_name: attempt-01
|
66 |
+
wandb_log_model:
|
67 |
+
|
68 |
+
gradient_accumulation_steps: 2
|
69 |
+
micro_batch_size: 2
|
70 |
+
num_epochs: 4
|
71 |
+
optimizer: adamw_bnb_8bit
|
72 |
+
lr_scheduler: cosine
|
73 |
+
learning_rate: 0.00001
|
74 |
+
|
75 |
+
train_on_inputs: false
|
76 |
+
group_by_length: false
|
77 |
+
bf16: auto
|
78 |
+
fp16:
|
79 |
+
tf32: false
|
80 |
+
|
81 |
+
gradient_checkpointing: true
|
82 |
+
early_stopping_patience:
|
83 |
+
auto_resume_from_checkpoints: true
|
84 |
+
local_rank:
|
85 |
+
logging_steps: 1
|
86 |
+
xformers_attention:
|
87 |
+
flash_attention: true
|
88 |
+
|
89 |
+
warmup_steps: 10
|
90 |
+
evals_per_epoch:
|
91 |
+
eval_table_size:
|
92 |
+
eval_max_new_tokens:
|
93 |
+
saves_per_epoch: 1
|
94 |
+
debug:
|
95 |
+
deepspeed: deepspeed_configs/zero3_bf16.json
|
96 |
+
weight_decay: 0.001
|
97 |
+
fsdp:
|
98 |
+
fsdp_config:
|
99 |
+
special_tokens:
|
100 |
+
pad_token: <pad>
|