# 学習のベースモデルに関する設定
base_model: google/gemma-2-2b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

# 学習後のモデルのHFへのアップロードに関する設定
hub_model_id: Aratako/gemma-2-2b-axolotl-sft-v1.0
hub_strategy: "end"
push_dataset_to_hub:
hf_use_auth_token: true

# Liger Kernelの設定（学習の軽量・高速化）
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

# 量子化に関する設定
load_in_8bit: false
load_in_4bit: true

# SFTに利用するchat templateの設定
chat_template: gemma

# 学習データセットの前処理に関する設定
datasets:
  - path: kanhatakeyama/ramdom-to-fixed-multiturn-Calm3
    split: 20240806filtered[0:10000]
    type: chat_template
    field_messages: messages
    message_field_role: role
    message_field_content: content
  - path: llm-jp/magpie-sft-v1.0
    split: train[0:10000]
    type: chat_template
    field_messages: conversations
    message_field_role: role
    message_field_content: content
  - path: Aratako/magpie-qwen2.5-32b-reasoning-100k-formatted
    split: train[0:10000]
    type: chat_template
    field_messages: conversations
    message_field_role: role
    message_field_content: content

# データセット、モデルの出力先に関する設定
shuffle_merged_datasets: true
dataset_prepared_path: /workspace/data/sft-data
output_dir: /workspace/data/models/gemma-2-2b-axolotl-sft-v1.0

# valid datasetのサイズ
val_set_size: 0.05

# LoRAに関する設定（フルファインチューニングしたい場合は全て空欄にする）
adapter: qlora
lora_model_dir:
lora_r: 16
lora_alpha: 32
lora_dropout: 0.05
lora_target_linear: true
lora_fan_in_fan_out:

# wandbに関する設定
wandb_project: axolotl
wandb_entity: aratako-lm
wandb_watch:
wandb_name: sft-lora-1
wandb_log_model:

# 学習に関する様々な設定
sequence_len: 4096
sample_packing: true
eval_sample_packing: false
pad_to_sequence_len: true

gradient_accumulation_steps: 16
micro_batch_size: 1
num_epochs: 1
optimizer: paged_adamw_8bit
lr_scheduler: cosine
cosine_min_lr_ratio: 0.1
learning_rate: 3e-4

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: false
early_stopping_patience:
auto_resume_from_checkpoints: true
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

save_strategy: steps
save_steps: 50
save_total_limit: 2

warmup_steps: 10
eval_steps: 50
eval_batch_size: 1
eval_table_size:
eval_max_new_tokens:
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16.json
weight_decay: 0.01
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

gemma-2-2b-axolotl-sft-v1.0

This model is a fine-tuned version of google/gemma-2-2b on the kanhatakeyama/ramdom-to-fixed-multiturn-Calm3, the llm-jp/magpie-sft-v1.0 and the Aratako/magpie-qwen2.5-32b-reasoning-100k-formatted datasets. It achieves the following results on the evaluation set:

Loss: 1.3378

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: 0.0003
train_batch_size: 1
eval_batch_size: 1
seed: 42
distributed_type: multi-GPU
num_devices: 2
gradient_accumulation_steps: 16
total_train_batch_size: 32
total_eval_batch_size: 2
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: 10
num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss
1.3251	0.3726	50	1.3855
1.3015	0.7452	100	1.3378

Framework versions

PEFT 0.14.0
Transformers 4.46.3
Pytorch 2.5.1+cu124
Datasets 3.1.0
Tokenizers 0.20.3