RichardErkhov/CorticalStack_-_mistral-7b-dolphin-sft-gguf

Quantization made by Richard Erkhov.

Github

Discord

Request more models

mistral-7b-dolphin-sft - GGUF

• Model creator: https://huggingface.co/CorticalStack/
• Original model: https://huggingface.co/CorticalStack/mistral-7b-dolphin-sft/

Name	Quant method	Size
mistral-7b-dolphin-sft.Q2_K.gguf	Q2_K	2.53GB
mistral-7b-dolphin-sft.IQ3_XS.gguf	IQ3_XS	2.81GB
mistral-7b-dolphin-sft.IQ3_S.gguf	IQ3_S	2.96GB
mistral-7b-dolphin-sft.Q3_K_S.gguf	Q3_K_S	2.95GB
mistral-7b-dolphin-sft.IQ3_M.gguf	IQ3_M	3.06GB
mistral-7b-dolphin-sft.Q3_K.gguf	Q3_K	3.28GB
mistral-7b-dolphin-sft.Q3_K_M.gguf	Q3_K_M	3.28GB
mistral-7b-dolphin-sft.Q3_K_L.gguf	Q3_K_L	3.56GB
mistral-7b-dolphin-sft.IQ4_XS.gguf	IQ4_XS	3.67GB
mistral-7b-dolphin-sft.Q4_0.gguf	Q4_0	3.83GB
mistral-7b-dolphin-sft.IQ4_NL.gguf	IQ4_NL	3.87GB
mistral-7b-dolphin-sft.Q4_K_S.gguf	Q4_K_S	3.86GB
mistral-7b-dolphin-sft.Q4_K.gguf	Q4_K	4.07GB
mistral-7b-dolphin-sft.Q4_K_M.gguf	Q4_K_M	4.07GB
mistral-7b-dolphin-sft.Q4_1.gguf	Q4_1	4.24GB
mistral-7b-dolphin-sft.Q5_0.gguf	Q5_0	4.65GB
mistral-7b-dolphin-sft.Q5_K_S.gguf	Q5_K_S	4.65GB
mistral-7b-dolphin-sft.Q5_K.gguf	Q5_K	4.78GB
mistral-7b-dolphin-sft.Q5_K_M.gguf	Q5_K_M	4.78GB
mistral-7b-dolphin-sft.Q5_1.gguf	Q5_1	5.07GB
mistral-7b-dolphin-sft.Q6_K.gguf	Q6_K	5.53GB
mistral-7b-dolphin-sft.Q8_0.gguf	Q8_0	7.17GB

Original model description:

license: apache-2.0

mistral-7b-dolphin-sft

mistral-7b-dolphin-sft is an SFT fine-tuned version of unsloth/mistral-7b-bnb-4bit using the cognitivecomputations/dolphin dataset.

Fine-tuning configuration

LoRA

• r: 256
• LoRA alpha: 128
• LoRA dropout: 0.0

Training arguments

• Epochs: 1
• Batch size: 4
• Gradient accumulation steps: 6
• Optimizer: adamw_torch_fused
• Max steps: 100
• Learning rate: 0.0002
• Weight decay: 0.1
• Learning rate scheduler type: linear
• Max seq length: 2048
• 4-bit bnb: True

Trained with Unsloth and Huggingface's TRL library.