Quantization made by Richard Erkhov.
zephyr-7b-sft-full - GGUF
- Model creator: https://huggingface.co/alignment-handbook/
- Original model: https://huggingface.co/alignment-handbook/zephyr-7b-sft-full/
Name | Quant method | Size |
---|---|---|
zephyr-7b-sft-full.Q2_K.gguf | Q2_K | 2.53GB |
zephyr-7b-sft-full.IQ3_XS.gguf | IQ3_XS | 2.81GB |
zephyr-7b-sft-full.IQ3_S.gguf | IQ3_S | 2.96GB |
zephyr-7b-sft-full.Q3_K_S.gguf | Q3_K_S | 2.95GB |
zephyr-7b-sft-full.IQ3_M.gguf | IQ3_M | 3.06GB |
zephyr-7b-sft-full.Q3_K.gguf | Q3_K | 3.28GB |
zephyr-7b-sft-full.Q3_K_M.gguf | Q3_K_M | 3.28GB |
zephyr-7b-sft-full.Q3_K_L.gguf | Q3_K_L | 3.56GB |
zephyr-7b-sft-full.IQ4_XS.gguf | IQ4_XS | 3.67GB |
zephyr-7b-sft-full.Q4_0.gguf | Q4_0 | 3.83GB |
zephyr-7b-sft-full.IQ4_NL.gguf | IQ4_NL | 3.87GB |
zephyr-7b-sft-full.Q4_K_S.gguf | Q4_K_S | 3.86GB |
zephyr-7b-sft-full.Q4_K.gguf | Q4_K | 4.07GB |
zephyr-7b-sft-full.Q4_K_M.gguf | Q4_K_M | 4.07GB |
zephyr-7b-sft-full.Q4_1.gguf | Q4_1 | 4.24GB |
zephyr-7b-sft-full.Q5_0.gguf | Q5_0 | 4.65GB |
zephyr-7b-sft-full.Q5_K_S.gguf | Q5_K_S | 4.65GB |
zephyr-7b-sft-full.Q5_K.gguf | Q5_K | 4.78GB |
zephyr-7b-sft-full.Q5_K_M.gguf | Q5_K_M | 4.78GB |
zephyr-7b-sft-full.Q5_1.gguf | Q5_1 | 5.07GB |
zephyr-7b-sft-full.Q6_K.gguf | Q6_K | 5.53GB |
Original model description:
license: apache-2.0 base_model: mistralai/Mistral-7B-v0.1 tags: - alignment-handbook - generated_from_trainer - trl - sft - generated_from_trainer datasets: - HuggingFaceH4/ultrachat_200k model-index: - name: zephyr-7b-sft-full results: []
zephyr-7b-sft-full
This model is a fine-tuned version of mistralai/Mistral-7B-v0.1 on the HuggingFaceH4/ultrachat_200k dataset. It achieves the following results on the evaluation set:
- Loss: 0.9353
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: 2e-05
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- distributed_type: multi-GPU
- num_devices: 8
- total_train_batch_size: 128
- total_eval_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 1
Training results
Training Loss | Epoch | Step | Validation Loss |
---|---|---|---|
0.9075 | 1.0 | 1090 | 0.9353 |
Framework versions
- Transformers 4.36.2
- Pytorch 2.1.2+cu121
- Datasets 2.14.6
- Tokenizers 0.15.0