ABX-AI's picture
Update README.md
b9f2a4f verified
metadata
license: other
library_name: transformers
tags:
  - mergekit
  - merge
  - mistral
  - not-for-all-audiences
base_model:
  - ABX-AI/Cerebral-Infinity-7B
  - ABX-AI/Starfinite-Laymospice-v2-7B
model-index:
  - name: Quantum-Citrus-9B
    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: 65.19
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=ABX-AI/Quantum-Citrus-9B
          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: 84.75
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=ABX-AI/Quantum-Citrus-9B
          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: 64.58
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=ABX-AI/Quantum-Citrus-9B
          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: 55.96
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=ABX-AI/Quantum-Citrus-9B
          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: 79.4
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=ABX-AI/Quantum-Citrus-9B
          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: 50.57
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=ABX-AI/Quantum-Citrus-9B
          name: Open LLM Leaderboard

GGUF / IQ / Imatrix for Quantum-Citrus-9B

image/png

Why Importance Matrix?

Importance Matrix, at least based on my testing, has shown to improve the output and performance of "IQ"-type quantizations, where the compression becomes quite heavy. The Imatrix performs a calibration, using a provided dataset. Testing has shown that semi-randomized data can help perserve more important segments as the compression is applied.

Related discussions in Github: [1] [2]

The imatrix.txt file that I used contains general, semi-random data, with some custom kink.

Quantum-Citrus-9B

This merge is another attempt at making and intelligent, refined and unaligned model.

Based on my tests so far, it has accomplished the goals, and I am continuing to experiment with my interactions with it.

It includes previous merges of Starling, Cerebrum, LemonadeRP, InfinityRP, and deep down has a base of layla v0.1, as I am not that happy with the result form using v0.2.

The model is intended for fictional storytelling and roleplaying and may not be intended for all audences.

Merge Details

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

Merge Method

This model was merged using the passthrough merge method.

Models Merged

The following models were included in the merge:

Configuration

The following YAML configuration was used to produce this model:

slices:
  - sources:
      - model: ABX-AI/Cerebral-Infinity-7B
        layer_range: [0, 20]
  - sources:
      - model: ABX-AI/Starfinite-Laymospice-v2-7B
        layer_range: [12, 32]
merge_method: passthrough
dtype: float16

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric Value
Avg. 66.74
AI2 Reasoning Challenge (25-Shot) 65.19
HellaSwag (10-Shot) 84.75
MMLU (5-Shot) 64.58
TruthfulQA (0-shot) 55.96
Winogrande (5-shot) 79.40
GSM8k (5-shot) 50.57