AlexHung29629/Qwen2.5-Omni-7B-Reasoning

See axolotl config

axolotl version: 0.8.1

base_model: AlexHung29629/Qwen-2.5-Omni-ThinkerTextModel
model_type: AutoModelForCausalLM
trust_remote_code: true

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_glu_activation: true
liger_fused_linear_cross_entropy: true

datasets:
  - path: /mnt/shared/twsc/alex/qwen/s1_claude.jsonl
    type: chat_template
    field_messages: conversations
    message_property_mappings:
      role: role
      content: content
    roles_to_train:
    train_on_eos:

dataset_prepared_path: ./sft_dataprep/
val_set_size: 0
output_dir: ./placeholder_sft/
shuffle_merged_datasets: false

sequence_len: 20000
sample_packing: true
eval_sample_packing: false
pad_to_sequence_len: true

wandb_project: Reasoning_TP1_2025
wandb_entity:
wandb_watch:
wandb_name: Qwen-2.5-Omni-Reasoning-250422_sft
wandb_log_model:

gradient_accumulation_steps: 2
micro_batch_size: 1
num_epochs: 5
optimizer: adamw_torch
lr_scheduler: cosine
learning_rate: 1e-5
max_grad_norm: 1.0

adam_beta1: 0.9
adam_beta2: 0.95
adam_epsilon: 1e-8

bf16: true
tf32: false

gradient_checkpointing: true
gradient_checkpointing_kwargs:
  use_reentrant: true
logging_steps: 1
flash_attention: true
xformers_attention: false
sdp_attention: false

warmup_ratio: 0.05
saves_per_epoch: 1
save_total_limit: 5
weight_decay: 0.1
deepspeed: /mnt/shared/twsc/alex/reasoning/zero3_bf16.json
special_tokens:
  pad_token: "<pad>"
tokens:
  - <think>
  - </think>
seed: 42

placeholder_sft/

This model is a fine-tuned version of AlexHung29629/Qwen-2.5-Omni-ThinkerTextModel on the /mnt/shared/twsc/alex/qwen/s1_claude.jsonl 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: 1e-05
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• distributed_type: multi-GPU
• num_devices: 32
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• total_eval_batch_size: 32
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.95) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 3
• num_epochs: 5.0

Training results

Framework versions

• Transformers 4.52.0.dev0
• Pytorch 2.6.0+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1