v3c_mistral_lora

This model is a fine-tuned version of peiyi9979/math-shepherd-mistral-7b-prm on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.2939
• Accuracy: 0.8636
• Precision: 0.8421
• Recall: 0.6324
• F1: 0.7223

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: 2e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 765837
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• total_eval_batch_size: 32
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
No log	0	0	0.6026	0.7339	0.6	0.1542	0.2453
0.6391	0.0248	20	0.5954	0.7361	0.6119	0.1621	0.2562
0.5891	0.0495	40	0.5570	0.7550	0.6176	0.3320	0.4319
0.4606	0.0743	60	0.4962	0.7794	0.6667	0.4269	0.5205
0.4229	0.0990	80	0.4433	0.7905	0.6649	0.5099	0.5772
0.3836	0.1238	100	0.4297	0.8160	0.7605	0.5020	0.6048
0.3363	0.1485	120	0.3676	0.8381	0.7892	0.5771	0.6667
0.2483	0.1733	140	0.3537	0.8404	0.8225	0.5494	0.6588
0.2803	0.1980	160	0.3468	0.8415	0.8481	0.5296	0.6521
0.2782	0.2228	180	0.3493	0.8237	0.8310	0.4664	0.5975
0.2174	0.2475	200	0.3329	0.8492	0.8232	0.5889	0.6866
0.2965	0.2723	220	0.3314	0.8448	0.8343	0.5573	0.6682
0.2379	0.2970	240	0.3736	0.8149	0.8468	0.4150	0.5570
0.1587	0.3218	260	0.3315	0.8404	0.8609	0.5138	0.6436
0.1769	0.3465	280	0.3329	0.8370	0.8313	0.5257	0.6441
0.1984	0.3713	300	0.3211	0.8537	0.8712	0.5613	0.6827
0.2109	0.3960	320	0.3064	0.8570	0.8333	0.6126	0.7062
0.1961	0.4208	340	0.3035	0.8625	0.8413	0.6285	0.7195
0.2369	0.4455	360	0.2959	0.8747	0.8365	0.6877	0.7549
0.2355	0.4703	380	0.3176	0.8537	0.8380	0.5929	0.6944
0.1538	0.4950	400	0.3098	0.8503	0.8554	0.5613	0.6778
0.2261	0.5198	420	0.2964	0.8659	0.8235	0.6640	0.7352
0.1894	0.5446	440	0.3085	0.8625	0.8772	0.5929	0.7075
0.2089	0.5693	460	0.3103	0.8592	0.8621	0.5929	0.7026
0.225	0.5941	480	0.2933	0.8670	0.8519	0.6364	0.7285
0.2837	0.6188	500	0.2955	0.8636	0.8283	0.6482	0.7273
0.2046	0.6436	520	0.2943	0.8647	0.8429	0.6364	0.7252
0.1548	0.6683	540	0.3003	0.8636	0.8421	0.6324	0.7223
0.1626	0.6931	560	0.2982	0.8625	0.8603	0.6087	0.7130
0.2065	0.7178	580	0.2877	0.8636	0.8186	0.6601	0.7309
0.1423	0.7426	600	0.3031	0.8603	0.8757	0.5850	0.7014
0.1743	0.7673	620	0.2920	0.8659	0.8511	0.6324	0.7256
0.1281	0.7921	640	0.2912	0.8659	0.8474	0.6364	0.7269
0.1879	0.8168	660	0.2938	0.8625	0.8449	0.6245	0.7182
0.1741	0.8416	680	0.2965	0.8625	0.8486	0.6206	0.7169
0.1429	0.8663	700	0.2911	0.8647	0.8359	0.6443	0.7277
0.2218	0.8911	720	0.2950	0.8625	0.8449	0.6245	0.7182
0.1608	0.9158	740	0.2995	0.8603	0.8508	0.6087	0.7097
0.2056	0.9406	760	0.2967	0.8592	0.8424	0.6126	0.7094
0.2127	0.9653	780	0.2944	0.8625	0.8413	0.6285	0.7195
0.2252	0.9901	800	0.2939	0.8636	0.8421	0.6324	0.7223

Framework versions

• PEFT 0.13.2
• Transformers 4.46.0
• Pytorch 2.5.1+cu124
• Datasets 3.1.0
• Tokenizers 0.20.3