README.md

metadata

language:
  - en
license: apache-2.0
library_name: transformers
tags:
  - mergekit
  - merge
base_model:
  - uukuguy/speechless-code-mistral-7b-v1.0
  - upaya07/Arithmo2-Mistral-7B
pipeline_tag: text-generation
model-index:
  - name: sethuiyer/CodeCalc-Mistral-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: 61.95
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=sethuiyer/CodeCalc-Mistral-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: 83.64
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=sethuiyer/CodeCalc-Mistral-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: 62.78
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=sethuiyer/CodeCalc-Mistral-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: 47.49
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=sethuiyer/CodeCalc-Mistral-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: 78.3
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=sethuiyer/CodeCalc-Mistral-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: 63.53
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=sethuiyer/CodeCalc-Mistral-7B
          name: Open LLM Leaderboard

CodeCalc-Mistral-7B

Configuration

The following YAML configuration was used to produce this model:

base_model: uukuguy/speechless-code-mistral-7b-v1.0
dtype: bfloat16
merge_method: ties
models:
- model: uukuguy/speechless-code-mistral-7b-v1.0
- model: upaya07/Arithmo2-Mistral-7B
  parameters:
    density:  [0.25, 0.35, 0.45, 0.35, 0.25]
    weight: [0.1, 0.25, 0.5, 0.25, 0.1]
parameters:
  int8_mask: true

Evaluation

T	Model	Average	ARC	HellaSwag	MMLU	TruthfulQA	Winogrande	GSM8K
🔍	sethuiyer/CodeCalc-Mistral-7B	66.33	61.95	83.64	62.78	47.79	78.3	63.53
📉	uukuguy/speechless-code-mistral-7b-v1.0	63.6	61.18	83.77	63.4	47.9	78.37	47.01

The merge appears to be successful, especially considering the substantial improvement in the GSM8K benchmark while maintaining comparable performance on other metrics.

Usage

Alpaca Instruction Format and Divine Intellect preset.

You are an intelligent programming assistant.

### Instruction:
Implement a linked list in C++

### Response:

Preset:

temperature: 1.31
top_p: 0.14
repetition_penalty: 1.17
top_k: 49

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	66.33
AI2 Reasoning Challenge (25-Shot)	61.95
HellaSwag (10-Shot)	83.64
MMLU (5-Shot)	62.78
TruthfulQA (0-shot)	47.79
Winogrande (5-shot)	78.30
GSM8k (5-shot)	63.53