language:
- zh
- en
license: apache-2.0
datasets:
- Azure99/blossom-chat-v2
- Azure99/blossom-math-v3
- Azure99/blossom-wizard-v2
- Azure99/blossom-orca-v2
model-index:
- name: blossom-v4-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: 62.03
name: normalized accuracy
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Azure99/blossom-v4-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: 82.9
name: normalized accuracy
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Azure99/blossom-v4-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: 62.48
name: accuracy
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Azure99/blossom-v4-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: 53.84
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Azure99/blossom-v4-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: 77.27
name: accuracy
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Azure99/blossom-v4-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: 43.14
name: accuracy
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Azure99/blossom-v4-mistral-7b
name: Open LLM Leaderboard
BLOSSOM-v4-mistral-7b
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, 20K Math single-turn instruction datasets, training for 1 epoch; the second stage used 50K Blossom chat multi-turn dialogue dataset, and 2% 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-v4-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|:
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. | 63.61 |
| AI2 Reasoning Challenge (25-Shot) | 62.03 |
| HellaSwag (10-Shot) | 82.90 |
| MMLU (5-Shot) | 62.48 |
| TruthfulQA (0-shot) | 53.84 |
| Winogrande (5-shot) | 77.27 |
| GSM8k (5-shot) | 43.14 |