See axolotl config

axolotl version: 0.13.0.dev0

base_model: KaraKaraWitch/CavesOfQwen3-8b
hub_model_id: KaraKaraWitch/kane-wo-narashite

load_in_8bit: true
load_in_4bit: false


chat_template: chatml
datasets:
  - path: train.jsonl
    type: chat_template

    field_messages: conversation
    train_on_eos: all
    message_property_mappings:
      role: from
      content: content
dataset_prepared_path: last_run_prepared
val_set_size: 0.1
output_dir: lora-out

adapter: lora
lora_model_dir:

sequence_len: 4096
sample_packing: true


lora_r: 64
lora_alpha: 32
lora_dropout: 0.05
lora_target_linear: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

wandb_project: azure-edge
wandb_entity:
wandb_watch:
wandb_name: kane-wo-narashite-3
wandb_log_model:

gradient_accumulation_steps: 4
micro_batch_size: 2
num_epochs: 4
optimizer: adamw_bnb_8bit
lr_scheduler: cosine
learning_rate: 0.0002

bf16: auto
tf32: false

gradient_checkpointing: true
resume_from_checkpoint:
logging_steps: 1
flash_attention: true

loss_watchdog_threshold: 5.0
loss_watchdog_patience: 3

warmup_ratio: 0.1
evals_per_epoch: 1
saves_per_epoch: 4
weight_decay: 0.0
special_tokens:

# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

kane-wo-narashite

This model is a fine-tuned version of KaraKaraWitch/CavesOfQwen3-8b on the train.jsonl dataset. It achieves the following results on the evaluation set:

• Loss: 1.3720
• Memory/max Mem Active(gib): 27.79
• Memory/max Mem Allocated(gib): 27.79
• Memory/device Mem Reserved(gib): 32.76

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: 0.0002
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• distributed_type: multi-GPU
• num_devices: 2
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• total_eval_batch_size: 4
• optimizer: Use OptimizerNames.ADAMW_BNB 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: 296
• training_steps: 2964

Training results

Training Loss	Epoch	Step	Validation Loss	Mem Active(gib)	Mem Allocated(gib)	Mem Reserved(gib)
No log	0	0	1.6717	22.43	22.43	23.25
1.3952	0.9993	741	1.4237	27.79	27.79	32.76
1.3623	1.9980	1482	1.3892	27.79	27.79	32.76
1.2508	2.9966	2223	1.3717	27.79	27.79	32.76
0.9945	3.9953	2964	1.3720	27.79	27.79	32.76

Framework versions

• PEFT 0.17.0
• Transformers 4.55.2
• Pytorch 2.7.1+cu126
• Datasets 4.0.0
• Tokenizers 0.21.4