e-zorzi/Qwen2.5-VL-3B-Instruct-sft-v2

See axolotl config

axolotl version: 0.12.2

base_model: Qwen/Qwen2.5-VL-3B-Instruct
processor_type: AutoProcessor

# these 3 lines are needed for now to handle vision chat templates w images
skip_prepare_dataset: true
remove_unused_columns: false
sample_packing: false


chat_template: qwen2_vl
datasets:
  - path: e-zorzi/reasoning_distractors_chat
    type: chat_template
    split: train

dataset_prepared_path: last_run_prepared
val_set_size: 0.0
output_dir: ./outputs/out_qwen_fft_new

#load_in_8bit: True
#adapter: lora

lora_model_dir:

sequence_len: 2046 #8192
pad_to_sequence_len: false

#lora_r: 64
#lora_alpha: 16
#lora_dropout: 0.05
#lora_target_modules: 'model.language_model.layers.[\d]+.(mlp|cross_attn|self_attn).(up|down|gate|q|k|v|o)_proj'

wandb_project: axolotl_finetunes
wandb_entity: edo_vi
wandb_watch:
wandb_name:
wandb_log_model:

gradient_accumulation_steps: 4
micro_batch_size: 8
num_epochs: 1
optimizer: adamw_bnb_8bit
lr_scheduler: cosine
learning_rate: 0.0001 #0.0002

bf16: true
fp16:
tf32: true

gradient_checkpointing: true
logging_steps: 1
flash_attention: true
eager_attention:

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

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

outputs/out_qwen_fft_new

This model is a fine-tuned version of Qwen/Qwen2.5-VL-3B-Instruct on the e-zorzi/reasoning_distractors_chat dataset.

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: 0.0001
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• 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: 31
• training_steps: 310

Training results

Framework versions

• Transformers 4.57.1
• Pytorch 2.8.0+cu128
• Datasets 4.3.0
• Tokenizers 0.22.1