README.md

metadata

language:
  - en
license: apache-2.0
library_name: transformers
tags:
  - mergekit
  - merge
base_model:
  - Nondzu/Mistral-7B-Instruct-v0.2-code-ft
  - NousResearch/Nous-Hermes-2-Mistral-7B-DPO
  - cognitivecomputations/dolphin-2.6-mistral-7b-dpo-laser
  - eren23/ogno-monarch-jaskier-merge-7b-OH-PREF-DPO
model-index:
  - name: Gonzo-Chat-7B
    results:
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: AI2 Reasoning Challenge (25-Shot)
          type: ai2_arc
          config: ARC-Challenge
          split: test
          args:
            num_few_shot: 25
        metrics:
          - type: acc_norm
            value: 65.02
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Badgids/Gonzo-Chat-7B
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: HellaSwag (10-Shot)
          type: hellaswag
          split: validation
          args:
            num_few_shot: 10
        metrics:
          - type: acc_norm
            value: 85.4
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Badgids/Gonzo-Chat-7B
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: MMLU (5-Shot)
          type: cais/mmlu
          config: all
          split: test
          args:
            num_few_shot: 5
        metrics:
          - type: acc
            value: 63.75
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Badgids/Gonzo-Chat-7B
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: TruthfulQA (0-shot)
          type: truthful_qa
          config: multiple_choice
          split: validation
          args:
            num_few_shot: 0
        metrics:
          - type: mc2
            value: 60.23
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Badgids/Gonzo-Chat-7B
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: Winogrande (5-shot)
          type: winogrande
          config: winogrande_xl
          split: validation
          args:
            num_few_shot: 5
        metrics:
          - type: acc
            value: 77.74
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Badgids/Gonzo-Chat-7B
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: GSM8k (5-shot)
          type: gsm8k
          config: main
          split: test
          args:
            num_few_shot: 5
        metrics:
          - type: acc
            value: 47.61
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Badgids/Gonzo-Chat-7B
          name: Open LLM Leaderboard

Gonzo-Chat-7B

Gonzo-Chat-7B is a merged LLM based on Mistral v0.01 with a 8192 Context length that likes to chat, roleplay, work with agents, do some lite programming, and then beat the brakes off you in the back alley...

The BEST Open Source 7B Street Fighting LLM of 2024!!!

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	66.63
AI2 Reasoning Challenge (25-Shot)	65.02
HellaSwag (10-Shot)	85.40
MMLU (5-Shot)	63.75
TruthfulQA (0-shot)	60.23
Winogrande (5-shot)	77.74
GSM8k (5-shot)	47.61

LLM-Colosseum Results

All contestents fought using the same LLM-Colosseum default settings. Each contestant fought 25 rounds with every other contestant.

https://github.com/OpenGenerativeAI/llm-colosseum

Gonzo-Chat-7B .vs Mistral v0.2, Dolphon-Mistral v0.2, Deepseek-Coder-6.7b-instruct

This is a merge of pre-trained language models created using mergekit.

Merge Details

Merge Method

This model was merged using the DARE TIES merge method using eren23/ogno-monarch-jaskier-merge-7b-OH-PREF-DPO as a base.

Models Merged

The following models were included in the merge:

• Nondzu/Mistral-7B-Instruct-v0.2-code-ft
• NousResearch/Nous-Hermes-2-Mistral-7B-DPO
• cognitivecomputations/dolphin-2.6-mistral-7b-dpo-laser

Configuration

The following YAML configuration was used to produce this model:

models:
  - model: eren23/ogno-monarch-jaskier-merge-7b-OH-PREF-DPO
    # No parameters necessary for base model
  - model: cognitivecomputations/dolphin-2.6-mistral-7b-dpo-laser
    parameters:
      density: 0.53
      weight: 0.4
  - model:  NousResearch/Nous-Hermes-2-Mistral-7B-DPO
    parameters:
      density: 0.53
      weight: 0.3
  - model: Nondzu/Mistral-7B-Instruct-v0.2-code-ft
    parameters:
      density: 0.53
      weight: 0.3
merge_method: dare_ties
base_model: eren23/ogno-monarch-jaskier-merge-7b-OH-PREF-DPO
parameters:
  int8_mask: true
dtype: bfloat16