MentaLLaMA-chat-7B-PsyCourse-fold2

This model is a fine-tuned version of klyang/MentaLLaMA-chat-7B-hf on the course-train-fold2 dataset. It achieves the following results on the evaluation set:

• Loss: 0.0289

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
0.7519	0.0775	50	0.6389
0.1485	0.1550	100	0.1124
0.074	0.2326	150	0.0650
0.0655	0.3101	200	0.0619
0.0598	0.3876	250	0.0512
0.0414	0.4651	300	0.0449
0.0427	0.5426	350	0.0414
0.0471	0.6202	400	0.0387
0.0433	0.6977	450	0.0362
0.0432	0.7752	500	0.0353
0.0445	0.8527	550	0.0353
0.0529	0.9302	600	0.0353
0.0313	1.0078	650	0.0318
0.0301	1.0853	700	0.0322
0.0289	1.1628	750	0.0338
0.0267	1.2403	800	0.0314
0.0314	1.3178	850	0.0317
0.0382	1.3953	900	0.0327
0.0354	1.4729	950	0.0320
0.0265	1.5504	1000	0.0321
0.0301	1.6279	1050	0.0333
0.0262	1.7054	1100	0.0312
0.0273	1.7829	1150	0.0306
0.0283	1.8605	1200	0.0297
0.0381	1.9380	1250	0.0299
0.0207	2.0155	1300	0.0294
0.0163	2.0930	1350	0.0329
0.0236	2.1705	1400	0.0311
0.0191	2.2481	1450	0.0310
0.0243	2.3256	1500	0.0308
0.0165	2.4031	1550	0.0327
0.0224	2.4806	1600	0.0329
0.0289	2.5581	1650	0.0319
0.014	2.6357	1700	0.0316
0.0182	2.7132	1750	0.0334
0.0175	2.7907	1800	0.0298
0.0218	2.8682	1850	0.0297
0.018	2.9457	1900	0.0289
0.01	3.0233	1950	0.0309
0.0109	3.1008	2000	0.0338
0.0076	3.1783	2050	0.0347
0.0087	3.2558	2100	0.0358
0.0092	3.3333	2150	0.0323
0.0078	3.4109	2200	0.0331
0.0109	3.4884	2250	0.0356
0.0137	3.5659	2300	0.0360
0.013	3.6434	2350	0.0350
0.0133	3.7209	2400	0.0353
0.0068	3.7984	2450	0.0357
0.012	3.8760	2500	0.0348
0.0088	3.9535	2550	0.0344
0.0066	4.0310	2600	0.0346
0.0052	4.1085	2650	0.0361
0.008	4.1860	2700	0.0374
0.0062	4.2636	2750	0.0383
0.005	4.3411	2800	0.0386
0.004	4.4186	2850	0.0395
0.0075	4.4961	2900	0.0400
0.003	4.5736	2950	0.0402
0.0066	4.6512	3000	0.0405
0.005	4.7287	3050	0.0406
0.0067	4.8062	3100	0.0407
0.0067	4.8837	3150	0.0407
0.006	4.9612	3200	0.0407

Framework versions

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