See axolotl config

axolotl version: 0.6.0

datasets:
- path: phunguyen01/OpenR1-Math-220k-default-train-10k
  subset: default
  split: train
  type: chat_template
  field_messages: messages
  message_field_role: role
  message_field_content: content
  roles:
    system:
    - system
    user:
    - user
    assistant:
    - assistant
test_datasets: null
sequence_len: 8192
chat_template: qwen_25
base_model: Qwen/Qwen2.5-3B
micro_batch_size: 4
num_epochs: 2
learning_rate: 5.0e-06
output_dir: experiments/Qwen-3B-R1-Math-10k
dataset_prepared_path: experiments/Qwen-3B-R1-Math-10k/dataset_prepared
seed: 42
gradient_accumulation_steps: 1
gradient_checkpointing: true
flash_attention: true
train_on_inputs: false
sample_packing: true
eval_sample_packing: false
pad_to_sequence_len: true
bf16: auto
logging_steps: 5
lr_scheduler: cosine
warmup_ratio: 0.1
weight_decay: 0.01
save_strategy: epoch
evals_per_epoch: 1
wandb_project: i1
wandb_name: Qwen-3B-R1-Math-10k
hub_model_id: phunguyen01/Qwen-3B-R1-Math-10k
push_to_hub: true
special_tokens:
  eos_token: <|im_end|>
  additional_special_tokens: ["<think>", "</think>"]

deepspeed: zero2.json

Qwen-3B-R1-Math-10k

This model is a fine-tuned version of Qwen/Qwen2.5-3B on the phunguyen01/OpenR1-Math-220k-default-train-10k 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: 5e-06
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• total_train_batch_size: 32
• total_eval_batch_size: 32
• optimizer: Use adamw_hf 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: 49
• num_epochs: 2.0

Training results

Framework versions

• Transformers 4.48.3
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0