collinear-ai/math_reasoning_phi_c2

This is an open-source fine-tuned reasoning adapter of microsoft/Phi-3.5-mini-instruct, transformed into a math reasoning model using data curated from collinear-ai/R1-Distill-SFT-Curated.

See axolotl version

axolotl version: 0.5.0

Intended uses & limitations

Math-Reassoning

Training and evaluation data

Training data curated from collinear-ai/R1-Distill-SFT-Curated Evaluation data: HuggingFaceH4/MATH-500

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-06
• train_batch_size: 10
• eval_batch_size: 10
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 2
• total_train_batch_size: 160
• total_eval_batch_size: 80
• 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: 50
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.0003	1	0.6646
0.3174	0.3335	1247	0.3329
0.307	0.6670	2494	0.3169

Evaluation on Math500

Framework versions

• PEFT 0.13.2
• Transformers 4.46.1
• Pytorch 2.3.1+cu121
• Datasets 3.0.1
• Tokenizers 0.20.3