db806de803d4453718136c725f35c8e5

This model is a fine-tuned version of deepseek-ai/DeepSeek-R1-Distill-Qwen-7B on the dair-ai/emotion [split] dataset. It achieves the following results on the evaluation set:

• Loss: 1.6038
• Data Size: 1.0
• Epoch Runtime: 436.4765
• Accuracy: 0.9244
• F1 Macro: 0.8808

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-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• total_train_batch_size: 32
• total_eval_batch_size: 32
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Data Size	Epoch Runtime	Accuracy	F1 Macro
No log	0	0	30.9953	0	6.4117	0.1467	0.0590
No log	1	500	17.8548	0.0078	8.5870	0.2863	0.1749
No log	2	1000	6.8559	0.0156	20.2647	0.5202	0.2735
No log	3	1500	3.7588	0.0312	34.5336	0.6996	0.6123
No log	4	2000	1.9829	0.0625	57.4625	0.8432	0.7964
0.2265	5	2500	2.1728	0.125	84.4405	0.8548	0.8164
1.212	6	3000	1.3173	0.25	143.5163	0.8901	0.7394
0.1434	7	3500	1.3661	0.5	230.8958	0.9133	0.8638
0.7733	8.0	4000	0.6688	1.0	409.8183	0.9284	0.8743
0.4879	9.0	4500	0.6411	1.0	407.9665	0.9279	0.8902
0.593	10.0	5000	0.6788	1.0	411.5251	0.9304	0.8917
0.5484	11.0	5500	0.9786	1.0	436.9212	0.9219	0.8818
0.3847	12.0	6000	1.2340	1.0	458.0374	0.9204	0.8694
0.3118	13.0	6500	1.6038	1.0	436.4765	0.9244	0.8808

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu128
• Datasets 4.3.0
• Tokenizers 0.22.1