README.md

metadata

base_model: tiiuae/Falcon3-10B-Instruct
tags:
  - text-generation-inference
  - transformers
  - unsloth
  - llama
  - trl
license: apache-2.0
language:
  - en
pipeline_tag: text-generation
library_name: transformers
model-index:
  - name: RA_Reasoner
    results:
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: IFEval (0-Shot)
          type: HuggingFaceH4/ifeval
          args:
            num_few_shot: 0
        metrics:
          - type: inst_level_strict_acc and prompt_level_strict_acc
            value: 55.92
            name: strict accuracy
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=Daemontatox/RA_Reasoner
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: BBH (3-Shot)
          type: BBH
          args:
            num_few_shot: 3
        metrics:
          - type: acc_norm
            value: 43.07
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=Daemontatox/RA_Reasoner
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: MATH Lvl 5 (4-Shot)
          type: hendrycks/competition_math
          args:
            num_few_shot: 4
        metrics:
          - type: exact_match
            value: 20.09
            name: exact match
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=Daemontatox/RA_Reasoner
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: GPQA (0-shot)
          type: Idavidrein/gpqa
          args:
            num_few_shot: 0
        metrics:
          - type: acc_norm
            value: 10.85
            name: acc_norm
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=Daemontatox/RA_Reasoner
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: MuSR (0-shot)
          type: TAUR-Lab/MuSR
          args:
            num_few_shot: 0
        metrics:
          - type: acc_norm
            value: 7.51
            name: acc_norm
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=Daemontatox/RA_Reasoner
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: MMLU-PRO (5-shot)
          type: TIGER-Lab/MMLU-Pro
          config: main
          split: test
          args:
            num_few_shot: 5
        metrics:
          - type: acc
            value: 36.67
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=Daemontatox/RA_Reasoner
          name: Open LLM Leaderboard

Uploaded Model

Developed by: Daemontatox

License: Apache 2.0

Finetuned from model: tiiuae/Falcon3-10B-Instruct

This model was fine-tuned from the Falcon-10B-Instruct model. It was trained 2x faster with Unsloth and Hugging Face's TRL library.

This model is intended for text generation tasks, with a focus on reasoning capabilities and instruction following, similar to capabilities demonstrated by the ChatGPT-O1-Mini model.

Training Details

This model was fine-tuned with Unsloth and TRL, resulting in significant speed improvements during the training process. Details on specific fine-tuning data, parameters and methods will be added soon. The fine-tuning process has prioritized improving the model's reasoning abilities on various benchmarks.

Intended Use

This model is intended for research and development purposes related to text generation, instruction following, and complex reasoning tasks. It is suitable for applications that require a model capable of handling multi-step logical problems and understanding nuanced instructions.

Focus on Reasoning: The fine-tuning has been geared towards enhancing the model's ability to tackle reasoning challenges and logic-based tasks.

---

Open LLM Leaderboard Evaluation Results

Detailed results can be found here! Summarized results can be found here!

Metric	% Value
Avg.	29.02
IFEval (0-Shot)	55.92
BBH (3-Shot)	43.07
MATH Lvl 5 (4-Shot)	20.09
GPQA (0-shot)	10.85
MuSR (0-shot)	7.51
MMLU-PRO (5-shot)	36.67