File size: 14,626 Bytes
5d45a16 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 |
---
library_name: transformers
tags:
- axolotl
- generated_from_trainer
datasets:
- Aratako/Magpie-Tanuki-Qwen2.5-72B-Answered
- Aratako/Open-Platypus-Japanese-masked-formatted
- llm-jp/wizardlm8x22b-logical-math-coding-sft-ja
- kanhatakeyama/ramdom-to-fixed-multiturn-Calm3
- llm-jp/Synthetic-JP-EN-Coding-Dataset
- llm-jp/magpie-sft-v1.0
model-index:
- name: plamo-2-1b-gorilla-chat5
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
[<img src="https://raw.githubusercontent.com/axolotl-ai-cloud/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/axolotl-ai-cloud/axolotl)
<details><summary>See axolotl config</summary>
axolotl version: `0.7.0`
```yaml
# モデルの設定
base_model: /notebooks/plamo-2-1b-gorilla-chat2 # HuggingFace上のモデル名
model_type: AutoModelForCausalLM # モデルのロードに使用するクラス
tokenizer_type: AutoTokenizer # トークナイザのロードに使用するクラス
trust_remote_code: true # リモートのカスタムコードを信頼してモデルをロード
hub_model_id: zamagi/fft-1
hub_strategy: "end"
push_dataset_to_hub:
hf_use_auth_token: true
plugins:
- axolotl.integrations.liger.LigerPlugin
liger_cross_entropy: false
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true
# 8bit/4bit設定(8bitモードでメモリ削減)
load_in_8bit: false #f # 8bit量子化されたモデルをロード
load_in_4bit: false # 4bit量子化は使用しない
strict: false # 重みの厳密な一致を要求しない(追加トークン等がある場合に許容)
chat_template: tokenizer_default
# データセットの設定
datasets:
- path: Aratako/Magpie-Tanuki-Qwen2.5-72B-Answered
type: chat_template
field_messages: messages
message_property_mappings: # メッセージ内のプロパティ名のマッピング
role: role # 役割(ユーザー/システム/アシスタント)を示すフィールド
content: content # メッセージ内容を示すフィールド
roles_to_train: ["assistant"] # 学習対象とする役割(アシスタントの発話のみ学習)
train_on_eos: last
# - path: Aratako/magpie-qwen2.5-32b-reasoning-100k-formatted
# type: chat_template
# field_messages: conversations
# message_field_role: role
# message_field_content: content
# roles_to_train: ["assistant"] # 学習対象とする役割(アシスタントの発話のみ学習)
# train_on_eos: last
# - path: Aratako/magpie-reasoning-llama-nemotron-70b-100k-filtered
# type: chat_template
# field_messages: conversations
# message_field_role: role
# message_field_content: content
- path: Aratako/Open-Platypus-Japanese-masked-formatted
type: chat_template
field_messages: conversations
message_property_mappings: # メッセージ内のプロパティ名のマッピング
role: role # 役割(ユーザー/システム/アシスタント)を示すフィールド
content: content # メッセージ内容を示すフィールド
roles_to_train: ["assistant"] # 学習対象とする役割(アシスタントの発話のみ学習)
train_on_eos: last
- path: llm-jp/wizardlm8x22b-logical-math-coding-sft-ja
type: chat_template
field_messages: messages
message_property_mappings: # メッセージ内のプロパティ名のマッピング
role: role # 役割(ユーザー/システム/アシスタント)を示すフィールド
content: content # メッセージ内容を示すフィールド
roles_to_train: ["assistant"] # 学習対象とする役割(アシスタントの発話のみ学習)
train_on_eos: last
- path: kanhatakeyama/ramdom-to-fixed-multiturn-Calm3
split: 20240806filtered
type: chat_template
field_messages: messages
message_property_mappings: # メッセージ内のプロパティ名のマッピング
role: role # 役割(ユーザー/システム/アシスタント)を示すフィールド
content: content # メッセージ内容を示すフィールド
roles_to_train: ["assistant"] # 学習対象とする役割(アシスタントの発話のみ学習)
train_on_eos: last
# - path: Aratako/magpie-ultra-v0.1-formatted
# type: chat_template
# field_messages: conversations
# message_field_role: role
# message_field_content: content
# - path: Aratako/orca-agentinstruct-1M-v1-selected
# type: chat_template
# field_messages: messages
# message_field_role: role
# message_field_content: content
- path: llm-jp/Synthetic-JP-EN-Coding-Dataset
type: chat_template
field_messages: messages
message_property_mappings: # メッセージ内のプロパティ名のマッピング
role: role # 役割(ユーザー/システム/アシスタント)を示すフィールド
content: content # メッセージ内容を示すフィールド
roles_to_train: ["assistant"] # 学習対象とする役割(アシスタントの発話のみ学習)
train_on_eos: last
- path: llm-jp/magpie-sft-v1.0 # 使用するデータセット(Hugging Face上のデータセット名)
type: chat_template # 会話形式のデータセットを使用
field_messages: conversations # 会話データが格納されたフィールド名
message_property_mappings: # メッセージ内のプロパティ名のマッピング
role: role # 役割(ユーザー/システム/アシスタント)を示すフィールド
content: content # メッセージ内容を示すフィールド
roles_to_train: ["assistant"] # 学習対象とする役割(アシスタントの発話のみ学習)
train_on_eos: last
shuffle_merged_datasets: true
dataset_prepared_path: /notebooks/data/fft-data
val_set_size: 0.002
output_dir: /notebooks/data/27b-fft-out-1
dataset_keep_in_memory: false
gpu_memory_limit: 48GiB
sequence_len: 2048
sample_packing: true
eval_sample_packing: false
pad_to_sequence_len: true
adapter:
lora_model_dir:
lora_r:
lora_alpha:
lora_dropout:
lora_target_linear:
lora_fan_in_fan_out:
# トレーニングの設定
gradient_accumulation_steps: 4
micro_batch_size: 8
num_epochs: 2
optimizer: paged_adamw_8bit
lr_scheduler:
cosine_min_lr_ratio: 0.1
learning_rate: 0.00001
max_steps: 10000
train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false
#wandb: false
#wandb_project: 27b-fft
#wandb_entity: aratako-lm
#wandb_watch:
#wandb_name: attempt-01
#wandb_log_model:
gradient_checkpointing: true
early_stopping_patience:
auto_resume_from_checkpoints: true
local_rank:
logging_steps: 1
xformers_attention:
flash_attention:
save_strategy: steps
save_steps: 100
save_total_limit: 2
warmup_steps: 50
eval_steps: 100
eval_batch_size: 1
eval_table_size:
eval_max_new_tokens:
debug:
deepspeed: /notebooks/axolotl/deepspeed_configs/zero3_bf16.json
weight_decay: 0.01
fsdp:
fsdp_config:
# 出力の保存設定
output_dir: /notebooks/output/plamo-2-1b-gorilla-chat5 # チェックポイントや最終モデルの出力先ディレクトリ
hub_model_id: zamagi/plamo-2-1b-gorilla-chat5 # (オプション) Hugging Face Hubにアップロードする場合のリポジトリ名
```
</details><br>
# plamo-2-1b-gorilla-chat5
This model was trained from scratch on the Aratako/Magpie-Tanuki-Qwen2.5-72B-Answered, the Aratako/Open-Platypus-Japanese-masked-formatted, the llm-jp/wizardlm8x22b-logical-math-coding-sft-ja, the kanhatakeyama/ramdom-to-fixed-multiturn-Calm3, the llm-jp/Synthetic-JP-EN-Coding-Dataset and the llm-jp/magpie-sft-v1.0 datasets.
It achieves the following results on the evaluation set:
- Loss: 1.2854
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 1e-05
- train_batch_size: 8
- eval_batch_size: 1
- seed: 42
- distributed_type: multi-GPU
- gradient_accumulation_steps: 4
- total_train_batch_size: 32
- optimizer: Use OptimizerNames.PAGED_ADAMW_8BIT with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 50
- training_steps: 10000
### Training results
| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:------:|:-----:|:---------------:|
| 1.4277 | 0.0002 | 1 | 1.5568 |
| 1.3262 | 0.0196 | 100 | 1.4437 |
| 1.2695 | 0.0391 | 200 | 1.4289 |
| 1.4199 | 0.0587 | 300 | 1.4149 |
| 1.2383 | 0.0783 | 400 | 1.4073 |
| 1.418 | 0.0979 | 500 | 1.3987 |
| 1.2148 | 0.1174 | 600 | 1.3954 |
| 1.3301 | 0.1370 | 700 | 1.3906 |
| 1.3418 | 0.1566 | 800 | 1.3850 |
| 1.248 | 0.1762 | 900 | 1.3801 |
| 1.3027 | 0.1957 | 1000 | 1.3762 |
| 1.3965 | 0.2153 | 1100 | 1.3768 |
| 1.2422 | 0.2349 | 1200 | 1.3747 |
| 1.2969 | 0.2544 | 1300 | 1.3682 |
| 1.248 | 0.2740 | 1400 | 1.3629 |
| 1.3203 | 0.2936 | 1500 | 1.3582 |
| 1.2637 | 0.3132 | 1600 | 1.3576 |
| 1.3398 | 0.3327 | 1700 | 1.3559 |
| 1.1934 | 0.3523 | 1800 | 1.3508 |
| 1.1992 | 0.3719 | 1900 | 1.3525 |
| 1.1816 | 0.3914 | 2000 | 1.3475 |
| 1.1562 | 0.4110 | 2100 | 1.3441 |
| 1.373 | 0.4306 | 2200 | 1.3374 |
| 1.2188 | 0.4502 | 2300 | 1.3383 |
| 1.1738 | 0.4697 | 2400 | 1.3376 |
| 1.2344 | 0.4893 | 2500 | 1.3318 |
| 1.291 | 0.5089 | 2600 | 1.3289 |
| 1.2148 | 0.5285 | 2700 | 1.3254 |
| 1.248 | 0.5480 | 2800 | 1.3245 |
| 1.2988 | 0.5676 | 2900 | 1.3260 |
| 1.3359 | 0.5872 | 3000 | 1.3255 |
| 1.2109 | 0.6067 | 3100 | 1.3222 |
| 1.2656 | 0.6263 | 3200 | 1.3191 |
| 1.2109 | 0.6459 | 3300 | 1.3160 |
| 1.2676 | 0.6655 | 3400 | 1.3136 |
| 1.1426 | 0.6850 | 3500 | 1.3137 |
| 1.2422 | 0.7046 | 3600 | 1.3262 |
| 1.2188 | 0.7242 | 3700 | 1.3283 |
| 1.2891 | 0.7437 | 3800 | 1.3277 |
| 1.1758 | 0.7633 | 3900 | 1.3232 |
| 1.1846 | 0.7829 | 4000 | 1.3268 |
| 1.3418 | 0.8025 | 4100 | 1.3235 |
| 1.2812 | 0.8220 | 4200 | 1.3214 |
| 1.2793 | 0.8416 | 4300 | 1.3202 |
| 1.1758 | 0.8612 | 4400 | 1.3196 |
| 1.2188 | 0.8808 | 4500 | 1.3198 |
| 1.1719 | 0.9003 | 4600 | 1.3177 |
| 1.1738 | 0.9199 | 4700 | 1.3129 |
| 1.3555 | 0.9395 | 4800 | 1.3154 |
| 1.2207 | 0.9590 | 4900 | 1.3152 |
| 1.1445 | 0.9786 | 5000 | 1.3110 |
| 1.2891 | 0.9982 | 5100 | 1.3094 |
| 1.0527 | 1.0178 | 5200 | 1.3123 |
| 1.0527 | 1.0374 | 5300 | 1.3120 |
| 1.1777 | 1.0570 | 5400 | 1.3124 |
| 1.0879 | 1.0765 | 5500 | 1.3128 |
| 1.1836 | 1.0961 | 5600 | 1.3114 |
| 1.1406 | 1.1157 | 5700 | 1.3117 |
| 1.1152 | 1.1352 | 5800 | 1.3092 |
| 1.1387 | 1.1548 | 5900 | 1.3106 |
| 1.2715 | 1.1744 | 6000 | 1.3063 |
| 1.1855 | 1.1940 | 6100 | 1.3070 |
| 1.1895 | 1.2135 | 6200 | 1.3070 |
| 1.1309 | 1.2331 | 6300 | 1.3063 |
| 1.0918 | 1.2527 | 6400 | 1.3043 |
| 1.0977 | 1.2723 | 6500 | 1.3050 |
| 1.0332 | 1.2918 | 6600 | 1.3028 |
| 0.9697 | 1.3114 | 6700 | 1.3012 |
| 1.1504 | 1.3310 | 6800 | 1.3006 |
| 1.1152 | 1.3505 | 6900 | 1.3013 |
| 1.0127 | 1.3701 | 7000 | 1.2998 |
| 1.1387 | 1.3897 | 7100 | 1.2993 |
| 1.0664 | 1.4093 | 7200 | 1.2970 |
| 1.1299 | 1.4288 | 7300 | 1.2971 |
| 1.1406 | 1.4484 | 7400 | 1.2971 |
| 1.0684 | 1.4680 | 7500 | 1.2969 |
| 1.0938 | 1.4875 | 7600 | 1.2966 |
| 1.1221 | 1.5071 | 7700 | 1.2943 |
| 1.0771 | 1.5267 | 7800 | 1.2937 |
| 1.1211 | 1.5463 | 7900 | 1.2938 |
| 1.043 | 1.5658 | 8000 | 1.2941 |
| 1.0537 | 1.5854 | 8100 | 1.2924 |
| 1.0859 | 1.6050 | 8200 | 1.2918 |
| 1.1836 | 1.6246 | 8300 | 1.2911 |
| 1.2188 | 1.6441 | 8400 | 1.2906 |
| 1.0596 | 1.6637 | 8500 | 1.2912 |
| 1.041 | 1.6833 | 8600 | 1.2904 |
| 1.1367 | 1.7028 | 8700 | 1.2904 |
| 1.1006 | 1.7224 | 8800 | 1.2891 |
| 1.0996 | 1.7420 | 8900 | 1.2898 |
| 1.1387 | 1.7616 | 9000 | 1.2883 |
| 1.1543 | 1.7811 | 9100 | 1.2888 |
| 1.1328 | 1.8007 | 9200 | 1.2876 |
| 1.0801 | 1.8203 | 9300 | 1.2872 |
| 1.1855 | 1.8398 | 9400 | 1.2880 |
| 1.1113 | 1.8594 | 9500 | 1.2860 |
| 1.1289 | 1.8790 | 9600 | 1.2865 |
| 1.1543 | 1.8986 | 9700 | 1.2857 |
| 1.123 | 1.9181 | 9800 | 1.2856 |
| 1.0352 | 1.9377 | 9900 | 1.2857 |
| 0.9189 | 1.9573 | 10000 | 1.2854 |
### Framework versions
- Transformers 4.49.0
- Pytorch 2.5.1+cu124
- Datasets 3.2.0
- Tokenizers 0.21.1
|