CoolFace
Modelpublic

zamagi/plamo-2-1b-gorilla-chat5

sourceHugging Faceupdated 1y agoView on Hugging Face
0likes10downloads
Model Card

<!-- 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"/> <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
  • —trainbatchsize: 8
  • —evalbatchsize: 1
  • —seed: 42
  • —distributed_type: multi-GPU
  • —gradientaccumulationsteps: 4
  • —totaltrainbatch_size: 32
  • —optimizer: Use OptimizerNames.PAGEDADAMW8BIT with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
  • —lrschedulertype: cosine
  • —lrschedulerwarmup_steps: 50
  • —training_steps: 10000

Training results

Training LossEpochStepValidation Loss
1.42770.000211.5568
1.32620.01961001.4437
1.26950.03912001.4289
1.41990.05873001.4149
1.23830.07834001.4073
1.4180.09795001.3987
1.21480.11746001.3954
1.33010.13707001.3906
1.34180.15668001.3850
1.2480.17629001.3801
1.30270.195710001.3762
1.39650.215311001.3768
1.24220.234912001.3747
1.29690.254413001.3682
1.2480.274014001.3629
1.32030.293615001.3582
1.26370.313216001.3576
1.33980.332717001.3559
1.19340.352318001.3508
1.19920.371919001.3525
1.18160.391420001.3475
1.15620.411021001.3441
1.3730.430622001.3374
1.21880.450223001.3383
1.17380.469724001.3376
1.23440.489325001.3318
1.2910.508926001.3289
1.21480.528527001.3254
1.2480.548028001.3245
1.29880.567629001.3260
1.33590.587230001.3255
1.21090.606731001.3222
1.26560.626332001.3191
1.21090.645933001.3160
1.26760.665534001.3136
1.14260.685035001.3137
1.24220.704636001.3262
1.21880.724237001.3283
1.28910.743738001.3277
1.17580.763339001.3232
1.18460.782940001.3268
1.34180.802541001.3235
1.28120.822042001.3214
1.27930.841643001.3202
1.17580.861244001.3196
1.21880.880845001.3198
1.17190.900346001.3177
1.17380.919947001.3129
1.35550.939548001.3154
1.22070.959049001.3152
1.14450.978650001.3110
1.28910.998251001.3094
1.05271.017852001.3123
1.05271.037453001.3120
1.17771.057054001.3124
1.08791.076555001.3128
1.18361.096156001.3114
1.14061.115757001.3117
1.11521.135258001.3092
1.13871.154859001.3106
1.27151.174460001.3063
1.18551.194061001.3070
1.18951.213562001.3070
1.13091.233163001.3063
1.09181.252764001.3043
1.09771.272365001.3050
1.03321.291866001.3028
0.96971.311467001.3012
1.15041.331068001.3006
1.11521.350569001.3013
1.01271.370170001.2998
1.13871.389771001.2993
1.06641.409372001.2970
1.12991.428873001.2971
1.14061.448474001.2971
1.06841.468075001.2969
1.09381.487576001.2966
1.12211.507177001.2943
1.07711.526778001.2937
1.12111.546379001.2938
1.0431.565880001.2941
1.05371.585481001.2924
1.08591.605082001.2918
1.18361.624683001.2911
1.21881.644184001.2906
1.05961.663785001.2912
1.0411.683386001.2904
1.13671.702887001.2904
1.10061.722488001.2891
1.09961.742089001.2898
1.13871.761690001.2883
1.15431.781191001.2888
1.13281.800792001.2876
1.08011.820393001.2872
1.18551.839894001.2880
1.11131.859495001.2860
1.12891.879096001.2865
1.15431.898697001.2857
1.1231.918198001.2856
1.03521.937799001.2857
0.91891.9573100001.2854

Framework versions

  • —Transformers 4.49.0
  • —Pytorch 2.5.1+cu124
  • —Datasets 3.2.0
  • —Tokenizers 0.21.1