README.md

metadata

language:
  - zh
  - en
license: apache-2.0
datasets:
  - Azure99/blossom-chat-v1
  - Azure99/blossom-math-v2
  - Azure99/blossom-wizard-v1
  - Azure99/blossom-orca-v1
model-index:
  - name: blossom-v3-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: 60.49
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Azure99/blossom-v3-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: 81.9
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Azure99/blossom-v3-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: 61.35
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Azure99/blossom-v3-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: 50.31
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Azure99/blossom-v3-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: 76.95
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Azure99/blossom-v3-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: 46.7
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Azure99/blossom-v3-mistral-7b
          name: Open LLM Leaderboard

BLOSSOM-v3-mistral-7b

💻Github • 🚀Blossom Chat Demo

Introduction

Blossom is a conversational large language model, fine-tuned on the Blossom Orca/Wizard/Chat/Math mixed dataset based on the Mistral-7B-v0.1 pre-trained model. Blossom possesses robust general capabilities and context comprehension. Additionally, the high-quality Chinese and English datasets used for training have been made open source.

Training was conducted in two stages. The first stage used 100K Wizard, 100K Orca single-turn instruction datasets, training for 1 epoch; the second stage used a 2K Blossom math reasoning dataset, 50K Blossom chat multi-turn dialogue dataset, and 1% randomly sampled data from the first stage, training for 3 epochs.

Note: The Mistral-7B-v0.1 pre-trained model is somewhat lacking in Chinese knowledge, so for Chinese scenarios, it is recommended to use blossom-v3-baichuan2-7b.

Inference

Inference is performed in the form of dialogue continuation.

Single-turn dialogue

A chat between a human and an artificial intelligence bot. The bot gives helpful, detailed, and polite answers to the human's questions.
|Human|: hello
|Bot|: Hello! How can I assist you today?

Multi-turn dialogue

A chat between a human and an artificial intelligence bot. The bot gives helpful, detailed, and polite answers to the human's questions.
|Human|: hello
|Bot|: Hello! How can I assist you today?</s>
|Human|: Generate a random number using python
|Bot|: 

Note: At the end of the Bot's output in the historical conversation, append a </s>.

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	62.95
AI2 Reasoning Challenge (25-Shot)	60.49
HellaSwag (10-Shot)	81.90
MMLU (5-Shot)	61.35
TruthfulQA (0-shot)	50.31
Winogrande (5-shot)	76.95
GSM8k (5-shot)	46.70