MentaLLaMA-chat-7B-PsyCourse-fold10

This model is a fine-tuned version of klyang/MentaLLaMA-chat-7B-hf on the course-train-fold10 dataset. It achieves the following results on the evaluation set:

• Loss: 0.0297

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: 0.0001
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 16
• optimizer: Use 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: 5.0

Training results

Training Loss	Epoch	Step	Validation Loss
0.8796	0.0770	50	0.6149
0.1225	0.1539	100	0.1079
0.0896	0.2309	150	0.0615
0.0656	0.3078	200	0.0519
0.0599	0.3848	250	0.0487
0.0559	0.4618	300	0.0453
0.047	0.5387	350	0.0427
0.0386	0.6157	400	0.0398
0.0402	0.6926	450	0.0429
0.0525	0.7696	500	0.0384
0.0374	0.8466	550	0.0381
0.0479	0.9235	600	0.0336
0.0464	1.0005	650	0.0337
0.0291	1.0774	700	0.0351
0.0266	1.1544	750	0.0336
0.0268	1.2314	800	0.0323
0.0257	1.3083	850	0.0325
0.0311	1.3853	900	0.0323
0.0345	1.4622	950	0.0317
0.0287	1.5392	1000	0.0345
0.0333	1.6162	1050	0.0313
0.0278	1.6931	1100	0.0314
0.0239	1.7701	1150	0.0306
0.0346	1.8470	1200	0.0312
0.0289	1.9240	1250	0.0301
0.0308	2.0010	1300	0.0319
0.0176	2.0779	1350	0.0314
0.0192	2.1549	1400	0.0312
0.0157	2.2318	1450	0.0315
0.0227	2.3088	1500	0.0301
0.0211	2.3858	1550	0.0305
0.0192	2.4627	1600	0.0327
0.0215	2.5397	1650	0.0311
0.0235	2.6166	1700	0.0297
0.0149	2.6936	1750	0.0312
0.0206	2.7706	1800	0.0299
0.0154	2.8475	1850	0.0318
0.0188	2.9245	1900	0.0301
0.0198	3.0014	1950	0.0300
0.0117	3.0784	2000	0.0322
0.0103	3.1554	2050	0.0334
0.0158	3.2323	2100	0.0343
0.0116	3.3093	2150	0.0330
0.0117	3.3862	2200	0.0342
0.01	3.4632	2250	0.0345
0.0113	3.5402	2300	0.0345
0.0111	3.6171	2350	0.0342
0.0105	3.6941	2400	0.0351
0.0094	3.7710	2450	0.0365
0.0144	3.8480	2500	0.0337
0.0071	3.9250	2550	0.0341
0.0081	4.0019	2600	0.0339
0.0051	4.0789	2650	0.0355
0.0081	4.1558	2700	0.0364
0.0068	4.2328	2750	0.0382
0.0055	4.3098	2800	0.0389
0.0045	4.3867	2850	0.0386
0.0033	4.4637	2900	0.0386
0.0031	4.5406	2950	0.0391
0.0055	4.6176	3000	0.0393
0.0081	4.6946	3050	0.0395
0.0034	4.7715	3100	0.0396
0.0073	4.8485	3150	0.0396
0.0043	4.9254	3200	0.0396

Framework versions

• PEFT 0.12.0
• Transformers 4.46.1
• Pytorch 2.5.1+cu124
• Datasets 3.1.0
• Tokenizers 0.20.3