README.md

metadata

license: mit
base_model: HuggingFaceH4/zephyr-7b-beta
tags:
  - generated_from_trainer
model-index:
  - name: models
    results: []

models

This model is a fine-tuned version of HuggingFaceH4/zephyr-7b-beta on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.7368

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: 0.0004
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 100
• training_steps: 40
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.9828	0.02	1	1.0330
1.1538	0.03	2	1.0256
0.9734	0.05	3	1.0120
1.0574	0.07	4	0.9942
0.9797	0.09	5	0.9755
0.9399	0.1	6	0.9580
1.0294	0.12	7	0.9434
0.7965	0.14	8	0.9318
0.7741	0.16	9	0.9236
0.8252	0.17	10	0.9178
0.8478	0.19	11	0.9135
0.9641	0.21	12	0.9068
0.9073	0.22	13	0.8980
0.9682	0.24	14	0.8877
0.8794	0.26	15	0.8774
0.7602	0.28	16	0.8690
0.9019	0.29	17	0.8611
0.8619	0.31	18	0.8547
0.8195	0.33	19	0.8484
0.9562	0.34	20	0.8418
0.7822	0.36	21	0.8366
0.767	0.38	22	0.8308
0.9024	0.4	23	0.8242
0.8596	0.41	24	0.8183
0.8424	0.43	25	0.8123
0.7396	0.45	26	0.8059
0.7742	0.47	27	0.7999
0.7007	0.48	28	0.7943
0.6915	0.5	29	0.7890
0.7054	0.52	30	0.7836
0.7622	0.53	31	0.7785
0.6493	0.55	32	0.7720
0.6106	0.57	33	0.7650
0.7534	0.59	34	0.7583
0.7065	0.6	35	0.7532
0.8823	0.62	36	0.7472
0.7082	0.64	37	0.7424
0.7292	0.66	38	0.7405
0.8142	0.67	39	0.7390
0.6079	0.69	40	0.7368

Framework versions

• Transformers 4.36.0.dev0
• Pytorch 2.1.0+cu118
• Datasets 2.15.0
• Tokenizers 0.15.0