fblgit/una-cybertron-7b-v1-fp16

Model Card for una-cybertron-7b-v1 (UNA: Uniform Neural Alignment)

We strike back, introducing Cybertron 7B v1 a 7B MistralAI based model, best on it's series. Trained on SFT, DPO and UNA (Unified Neural Alignment) on multiple datasets. He scores 64.60+ on HF LeaderTests (without DROP for now).

Scoring #1 at 2 December 2023:

Model	Average	ARC (25-s)	HellaSwag (10-s)	MMLU (5-s)	TruthfulQA (MC) (0-s)	Winogrande (5-s)	GSM8K (5-s)
mistralai/Mistral-7B-v0.1	60.97	59.98	83.31	64.16	42.15	78.37	37.83
perlthoughts/Chupacabra-7B-v2	63.54	66.47	85.17	64.49	57.6	79.16	28.35
fblgit/una-cybertron-7b-v1	64.60	68.17	85.14	62.07	63.98	80.9	27.34

The model excels in mathematics, logic, reasoning, overall very smart.

Model Details

Adiestrated with UNA: Uniform Neural Alignment technique (paper going out soon).

Model Description

• Developed by: juanako.ai
• Author: Xavier M.
• Model type: MistralAI 7B
• Funded by Cybertron's H100's

Prompt

The model is very good, works well on almost any prompt but ChatML format and Alpaca System gets the best

<|im_start|>system
- You are a helpful assistant chatbot trained by MosaicML.
- You answer questions.
- You are excited to be able to help the user, but will refuse to do anything that could be considered harmful to the user.
- You are more than just an information source, you are also able to write poetry, short stories, and make jokes.<|im_end|>
<|im_start|>user
Explain QKV<|im_end|>
<|im_start|>assistant

### Assistant: I am StableVicuna, a large language model created by CarperAI. I am here to chat!

### Human: Explain QKV
### Assistant:

[Round <|round|>]
问：Explain QKV
答：

[Round <|round|>]
Question：Explain QKV
Answer：

Question：Explain QKV
Answer：

Evaluation

|    Tasks     |Version|Shots | Metric |Value |   |Stderr|
|--------------|-------|------|--------|-----:|---|-----:|
|arc_challenge |       | 25   |acc_norm|0.6817|±  |0.0136|
|truthfulqa_mc2|       | 0    |acc     |0.6398|±  |0.0151|
|hellaswag     |       | 10   |acc_norm|0.8492|±  |0.0036|
|winogrande    |       | 0    |acc     |0.809 |±  |0.011 |
|gsm8k         |       | 5    |acc     |0.2733|±  |0.0137|
|mmlu          |       | 5    |acc     |0.6207|±  |0.1230|
|              |average|      |acc     |0.6456|   |      |

|      Groups      |Version|Filter|n-shot|Metric|Value |   |Stderr|
|------------------|-------|------|-----:|------|-----:|---|-----:|
|mmlu              |N/A    |none  |     0|acc   |0.6207|_  |0.1230|
| - humanities     |N/A    |none  |     5|acc   |0.5675|_  |0.1125|
| - other          |N/A    |none  |     5|acc   |0.6933|_  |0.1108|
| - social_sciences|N/A    |none  |     5|acc   |0.7270|_  |0.0666|
| - stem           |N/A    |none  |     5|acc   |0.5249|_  |0.1311|

Framework versions

• Transformers 4.35.0-UNA
• Pytorch 2.1.0
• Datasets 2.14.6
• Tokenizers 0.14.1

Citations

If you find Cybertron, Juanako or any of our models useful, specially if you use it for your big brand.. cite please:

@misc{unacybertron7a,
  title={Cybertron: Uniform Neural Alignment}, 
  author={Xavier Murias},
  year={2023},
  publisher = {HuggingFace},
  journal = {HuggingFace repository},
  howpublished = {\url{https://huggingface.co/fblgit/una-cybertron-7b-v1}},
}

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	69.49
AI2 Reasoning Challenge (25-Shot)	68.43
HellaSwag (10-Shot)	85.42
MMLU (5-Shot)	63.34
TruthfulQA (0-shot)	63.28
Winogrande (5-shot)	81.37
GSM8k (5-shot)	55.12