RichardErkhov/EmbeddedLLM_-_Mistral-7B-Merge-02-v0-gguf

Quantization made by Richard Erkhov.

Github

Discord

Request more models

Mistral-7B-Merge-02-v0 - GGUF

• Model creator: https://huggingface.co/EmbeddedLLM/
• Original model: https://huggingface.co/EmbeddedLLM/Mistral-7B-Merge-02-v0/

Name	Quant method	Size
Mistral-7B-Merge-02-v0.Q2_K.gguf	Q2_K	2.53GB
Mistral-7B-Merge-02-v0.IQ3_XS.gguf	IQ3_XS	2.81GB
Mistral-7B-Merge-02-v0.IQ3_S.gguf	IQ3_S	2.96GB
Mistral-7B-Merge-02-v0.Q3_K_S.gguf	Q3_K_S	2.95GB
Mistral-7B-Merge-02-v0.IQ3_M.gguf	IQ3_M	3.06GB
Mistral-7B-Merge-02-v0.Q3_K.gguf	Q3_K	3.28GB
Mistral-7B-Merge-02-v0.Q3_K_M.gguf	Q3_K_M	3.28GB
Mistral-7B-Merge-02-v0.Q3_K_L.gguf	Q3_K_L	3.56GB
Mistral-7B-Merge-02-v0.IQ4_XS.gguf	IQ4_XS	3.67GB
Mistral-7B-Merge-02-v0.Q4_0.gguf	Q4_0	3.83GB
Mistral-7B-Merge-02-v0.IQ4_NL.gguf	IQ4_NL	3.87GB
Mistral-7B-Merge-02-v0.Q4_K_S.gguf	Q4_K_S	3.86GB
Mistral-7B-Merge-02-v0.Q4_K.gguf	Q4_K	4.07GB
Mistral-7B-Merge-02-v0.Q4_K_M.gguf	Q4_K_M	4.07GB
Mistral-7B-Merge-02-v0.Q4_1.gguf	Q4_1	4.24GB
Mistral-7B-Merge-02-v0.Q5_0.gguf	Q5_0	4.65GB
Mistral-7B-Merge-02-v0.Q5_K_S.gguf	Q5_K_S	4.65GB
Mistral-7B-Merge-02-v0.Q5_K.gguf	Q5_K	4.78GB
Mistral-7B-Merge-02-v0.Q5_K_M.gguf	Q5_K_M	4.78GB
Mistral-7B-Merge-02-v0.Q5_1.gguf	Q5_1	5.07GB
Mistral-7B-Merge-02-v0.Q6_K.gguf	Q6_K	5.53GB
Mistral-7B-Merge-02-v0.Q8_0.gguf	Q8_0	7.17GB

Original model description:

license: apache-2.0 language: - en tags: - merge base_model: - teknium/OpenHermes-2.5-Mistral-7B - Intel/neural-chat-7b-v3-3

Model Description

This is an experiment to compare merging 2 models using DARE TIES versus SLERP 🦙

We are mainly interested to compare against Weyaxi/OpenHermes-2.5-neural-chat-v3-3-Slerp

The 2 models involved in the merge as follows:

1. teknium/OpenHermes-2.5-Mistral-7B
2. Intel/neural-chat-7b-v3-3

• base model: mistralai/Mistral-7B-v0.1

The yaml config file for the merge is:

models:
  - model: mistralai/Mistral-7B-v0.1
    # no parameters necessary for base model
  - model: teknium/OpenHermes-2.5-Mistral-7B
    parameters:
      weight: 0.5
      density: 0.5
  - model: Intel/neural-chat-7b-v3-3
    parameters:
      weight: 0.5
      density: 0.5
merge_method: dare_ties
base_model: mistralai/Mistral-7B-v0.1
parameters:
  int8_mask: true
dtype: bfloat16

Open LLM Leaderboard

Note that with more tuning DARE TIES might achieve better results.

	DARE TIES	SLERP
Average	70.69	71.38
ARC	67.49	68.09
HellaSwag	85.78	86.2
MMLU	64.1	64.26
TruthfulQA	60.52	62.78
Winogrande	79.01	79.16
GSM8K	67.25	67.78