Edit model card

MedQA_L3_1000steps_1e5rate_SFT

This model is a fine-tuned version of meta-llama/Meta-Llama-3-8B-Instruct on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 0.3681

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-05
  • train_batch_size: 2
  • eval_batch_size: 1
  • seed: 42
  • gradient_accumulation_steps: 2
  • total_train_batch_size: 4
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: cosine
  • lr_scheduler_warmup_steps: 100
  • training_steps: 1000

Training results

Training Loss Epoch Step Validation Loss
0.4577 0.0489 50 0.5024
0.4969 0.0977 100 0.4876
0.4689 0.1466 150 0.4380
0.4891 0.1954 200 0.4313
0.424 0.2443 250 0.4275
0.4408 0.2931 300 0.4208
0.4124 0.3420 350 0.4160
0.4012 0.3908 400 0.4113
0.4305 0.4397 450 0.4285
0.4031 0.4885 500 0.3974
0.3863 0.5374 550 0.3916
0.3981 0.5862 600 0.3861
0.3705 0.6351 650 0.3810
0.3591 0.6839 700 0.3760
0.3642 0.7328 750 0.3722
0.3712 0.7816 800 0.3699
0.3893 0.8305 850 0.3686
0.3512 0.8793 900 0.3682
0.3546 0.9282 950 0.3681
0.3736 0.9770 1000 0.3681

Framework versions

  • Transformers 4.41.0
  • Pytorch 2.0.0+cu117
  • Datasets 2.19.1
  • Tokenizers 0.19.1
Downloads last month
2
Safetensors
Model size
8.03B params
Tensor type
FP16
·
Inference API
Input a message to start chatting with tsavage68/MedQA_L3_1000steps_1e5rate_SFT.
Model is too large to load in Inference API (serverless). To try the model, launch it on Inference Endpoints (dedicated) instead.

Finetuned from