Echo-IE-3B-v0.1

This model is a fine-tuned version of meta-llama/Llama-3.2-3B-Instruct on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.1664
• Rewards/chosen: -0.0370
• Rewards/rejected: -0.3280
• Rewards/accuracies: 1.0
• Rewards/margins: 0.2910
• Logps/rejected: -3.2803
• Logps/chosen: -0.3698
• Logits/rejected: 1.0091
• Logits/chosen: 0.9877
• Nll Loss: 0.1600
• Log Odds Ratio: -0.0425
• Log Odds Chosen: 4.2039

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: 6e-06
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 10
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen	Nll Loss	Log Odds Ratio	Log Odds Chosen
0.3327	1.0	44	0.3234	-0.0664	-0.1431	1.0	0.0767	-1.4309	-0.6641	0.4718	0.5480	0.2905	-0.2860	1.2590
0.2004	2.0	88	0.2283	-0.0488	-0.2296	1.0	0.1809	-2.2965	-0.4877	0.6711	0.7162	0.2142	-0.1075	2.7194
0.1661	3.0	132	0.1974	-0.0423	-0.2767	1.0	0.2344	-2.7672	-0.4230	0.8238	0.8408	0.1878	-0.0679	3.4301
0.1227	4.0	176	0.1813	-0.0392	-0.2999	1.0	0.2607	-2.9992	-0.3919	0.8916	0.8935	0.1734	-0.0541	3.7906
0.1434	5.0	220	0.1743	-0.0380	-0.3141	1.0	0.2762	-3.1414	-0.3799	0.9271	0.9167	0.1671	-0.0484	4.0032
0.0994	6.0	264	0.1697	-0.0373	-0.3202	1.0	0.2828	-3.2017	-0.3732	0.9822	0.9679	0.1629	-0.0453	4.0966
0.0896	7.0	308	0.1677	-0.0371	-0.3247	1.0	0.2876	-3.2469	-0.3706	0.9892	0.9698	0.1612	-0.0436	4.1599
0.1047	8.0	352	0.1666	-0.0370	-0.3268	1.0	0.2899	-3.2685	-0.3695	1.0025	0.9822	0.1602	-0.0429	4.1914
0.0979	9.0	396	0.1662	-0.0369	-0.3281	1.0	0.2911	-3.2808	-0.3694	1.0120	0.9910	0.1598	-0.0426	4.2063
0.0986	10.0	440	0.1664	-0.0370	-0.3280	1.0	0.2910	-3.2803	-0.3698	1.0091	0.9877	0.1600	-0.0425	4.2039

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.1+cu121
• Datasets 3.0.0
• Tokenizers 0.19.1