tinyllama-1.1b-chat-v1.0-ui-math-coding-dpo-2

This model is a fine-tuned version of TinyLlama/TinyLlama-1.1B-Chat-v1.0 on the data/ui_math and the data/ui_coding datasets. It achieves the following results on the evaluation set:

• Loss: 0.3362
• Rewards/chosen: -3.2340
• Rewards/rejected: -7.1489
• Rewards/accuracies: 0.875
• Rewards/margins: 3.9149
• Logps/rejected: -1058.2659
• Logps/chosen: -682.5323
• Logits/rejected: -1.3117
• Logits/chosen: -1.3352

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-07
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 4
• total_train_batch_size: 256
• total_eval_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.2293	1.2210	1000	0.3224	-3.3080	-7.0491	0.9062	3.7411	-1048.2871	-689.9403	-1.4398	-1.4575

Framework versions

• Transformers 4.44.1
• Pytorch 2.1.2+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1