htlou/backup_0202_llamafactory_Compcap_cooccur_0_60-llava-mistral

Compcap_cooccur_0_60

This model is a fine-tuned version of llava-hf/llava-v1.6-mistral-7b-hf on the Compcap_cooccur_0_60 dataset. It achieves the following results on the evaluation set:

• Loss: 0.8070

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
0.9196	0.2257	50	0.9317
0.8737	0.4515	100	0.8765
0.8557	0.6772	150	0.8530
0.8242	0.9029	200	0.8354
0.7715	1.1287	250	0.8278
0.7752	1.3544	300	0.8197
0.7359	1.5801	350	0.8135
0.7467	1.8059	400	0.8078
0.7024	2.0316	450	0.8078
0.6984	2.2573	500	0.8086
0.7063	2.4831	550	0.8080
0.6873	2.7088	600	0.8071
0.7067	2.9345	650	0.8070

Framework versions

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