htlou/backup_0202_llamafactory_CompCap-GPT4-llava-mistral

CompCap-GPT4

This model is a fine-tuned version of llava-hf/llava-v1.6-mistral-7b-hf on the CompCap-GPT4 dataset. It achieves the following results on the evaluation set:

• Loss: 0.7652

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: 1e-06
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 4
• total_train_batch_size: 256
• total_eval_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 10
• num_epochs: 3.0

Training results

Training Loss	Epoch	Step	Validation Loss
0.9184	0.1296	50	0.9209
0.8611	0.2592	100	0.8653
0.8257	0.3889	150	0.8396
0.8308	0.5185	200	0.8231
0.8168	0.6481	250	0.8109
0.8021	0.7777	300	0.8005
0.7741	0.9073	350	0.7929
0.7155	1.0369	400	0.7880
0.7322	1.1666	450	0.7837
0.7214	1.2962	500	0.7790
0.6936	1.4258	550	0.7753
0.7046	1.5554	600	0.7717
0.6967	1.6850	650	0.7690
0.7197	1.8146	700	0.7658
0.704	1.9443	750	0.7633
0.6546	2.0739	800	0.7677
0.651	2.2035	850	0.7673
0.6601	2.3331	900	0.7667
0.6669	2.4627	950	0.7662
0.6566	2.5924	1000	0.7657
0.6654	2.7220	1050	0.7655
0.6389	2.8516	1100	0.7653
0.6574	2.9812	1150	0.7653

Framework versions

• Transformers 4.45.2
• Pytorch 2.4.0+cu121
• Datasets 2.21.0
• Tokenizers 0.20.3