TencentARC_-_LLaMA-Pro-8B-gguf

RichardErkhov/TencentARC_-_LLaMA-Pro-8B-gguf

TencentARC_-_LLaMA-Pro-8B-gguf is a machine learning model from RichardErkhov. Use it for the machine learning task on the model card, and read the license before you ship it in a product.

Downloads · 30 days

161

3% of all-time downloads

All-time downloads

5.9K

Public

Repo size

94.8 GB

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.gguf110 GB · 100%

At a glance

Access: Public
Created: May 16, 2024
Updated: May 16, 2024
SHA: 18c13c15

Model card

From the Hugging Face model README

Quantization made by Richard Erkhov.

Github

Discord

Request more models

LLaMA-Pro-8B - GGUF

Model creator: https://huggingface.co/TencentARC/
Original model: https://huggingface.co/TencentARC/LLaMA-Pro-8B/

Name	Quant method	Size
LLaMA-Pro-8B.Q2_K.gguf	Q2_K	2.91GB
LLaMA-Pro-8B.IQ3_XS.gguf	IQ3_XS	3.22GB
LLaMA-Pro-8B.IQ3_S.gguf	IQ3_S	3.39GB
LLaMA-Pro-8B.Q3_K_S.gguf	Q3_K_S	3.39GB
LLaMA-Pro-8B.IQ3_M.gguf	IQ3_M	3.59GB
LLaMA-Pro-8B.Q3_K.gguf	Q3_K	3.8GB
LLaMA-Pro-8B.Q3_K_M.gguf	Q3_K_M	3.8GB
LLaMA-Pro-8B.Q3_K_L.gguf	Q3_K_L	4.15GB
LLaMA-Pro-8B.IQ4_XS.gguf	IQ4_XS	4.2GB
LLaMA-Pro-8B.Q4_0.gguf	Q4_0	4.41GB
LLaMA-Pro-8B.IQ4_NL.gguf	IQ4_NL	4.44GB
LLaMA-Pro-8B.Q4_K_S.gguf	Q4_K_S	4.45GB
LLaMA-Pro-8B.Q4_K.gguf	Q4_K	4.71GB
LLaMA-Pro-8B.Q4_K_M.gguf	Q4_K_M	4.71GB
LLaMA-Pro-8B.Q4_1.gguf	Q4_1	4.89GB
LLaMA-Pro-8B.Q5_0.gguf	Q5_0	5.37GB
LLaMA-Pro-8B.Q5_K_S.gguf	Q5_K_S	5.37GB
LLaMA-Pro-8B.Q5_K.gguf	Q5_K	5.52GB
LLaMA-Pro-8B.Q5_K_M.gguf	Q5_K_M	5.52GB
LLaMA-Pro-8B.Q5_1.gguf	Q5_1	5.85GB
LLaMA-Pro-8B.Q6_K.gguf	Q6_K	6.39GB
LLaMA-Pro-8B.Q8_0.gguf	Q8_0	8.27GB

Original model description:

license: llama2

LLaMA-Pro-8B Model Card

Model Description

LLaMA-Pro is a progressive version of the original LLaMA model, enhanced by the addition of Transformer blocks. It specializes in integrating both general language understanding and domain-specific knowledge, particularly in programming and mathematics.

Development and Training

Developed by Tencent's ARC Lab, LLaMA-Pro is an 8.3 billion parameter model. It's an expansion of LLaMA2-7B, further trained on code and math corpora totaling 80 billion tokens.

Intended Use

This model is designed for a wide range of NLP tasks, with a focus on programming, mathematics, and general language tasks. It suits scenarios requiring integration of natural and programming languages.

Performance

LLaMA-Pro demonstrates advanced performance across various benchmarks. It outperforms existing models in the LLaMA series in handling diverse tasks, showcasing its capability as an intelligent language agent.

Overall Performance on Languages, math and code tasks

Model	ARC	Hellaswag	MMLU	TruthfulQA	Winogrande	GSM8K	GSM8K-PoT	HumanEval	MBPP	Avg
LLAMA PRO (8B)	54.10	77.94	47.88	39.04	73.95	17.89	25.42	28.66	33.20	44.2
LLaMA2-7B	53.07	78.59	46.87	38.76	74.03	14.48	17.68	13.05	20.09	39.62
CodeLLaMA-7B	39.93	60.80	31.12	37.82	64.01	5.16	25.20	33.50	41.40	37.66
LLAMA PRO-INSTRUCT	52.30	76.88	52.57	48.80	72.53	43.59	55.61	44.51	37.88	53.8

Performance on GPT4 Evaluation

Model	MT Bench
Alpaca-13B	4.53
CodeLLaMA-7B-Instruct	5.71
Vicuna-7B	6.17
LLaMA2-7B-Chat	6.27
LLAMA PRO-INSTRUCT	6.32

Limitations

While LLaMA-Pro addresses some limitations of previous models in the series, it may still encounter challenges specific to highly specialized domains or tasks.

Ethical Considerations

Users should be aware of potential biases in the model and use it responsibly, considering its impact on various applications.

Created: May 16, 2024
Updated: May 16, 2024