Shenzhi Wang🌟 @ShenzhiWang_THU Twitter profile

shenzhi-wang/Gemma-2-9B-Chinese-Chat · Hugging Face

23

90

377

Shenzhi Wang🌟

@ShenzhiWang_THU

4 months

🔥Introducing Gemma-2-9B-Chinese-Chat: the 1st Gemma-2 model tailored for Chinese&English users, fine-tuned on >100K preference pairs! 🚀Our model excels in Chinese prompts, and shows improved logic, coding, math, and writing skills. More info: 🧵⬇️

shenzhi-wang/Llama3-8B-Chinese-Chat · Hugging Face

10

33

190

Shenzhi Wang🌟

@ShenzhiWang_THU

6 months

🔥We introduce Llama3-8B-Chinese-Chat v2.1! 🚀Compared to v1, the training dataset of v2.1 is 5x larger, and it exhibits significant enhancements, especially in roleplay, function calling, and math capabilities! 😆Don't miss out on our v2.1 model ⬇️

shenzhi-wang/Gemma-2-27B-Chinese-Chat · Hugging Face

Shenzhi Wang🌟

@ShenzhiWang_THU

6 months

🔥Introducing Llama3-8B-Chinese-Chat, the 1st llama3 model finetuned on English-Chinese datasets via ORPO. 🚀Our model consistently produces better responses for Chinese prompts than the Llama-3-8B-Insturct, and excels in logic, coding, math, and writing.

23

90

377

4

22

120

Shenzhi Wang🌟

@ShenzhiWang_THU

4 months

🔥 After the 9B model, we present Gemma-2-27B-Chinese-Chat: the 1st Gemma2 27B model optimized for Chinese&English, finetuned on >100K preference pairs! 🚀 Our model excels in Chinese, with improved logic, coding, math, and writing skills. More info:🧵⬇️

shenzhi-wang/Llama3-70B-Chinese-Chat · Hugging Face

6

22

121

Shenzhi Wang🌟

@ShenzhiWang_THU

5 months

🚀 Presenting Llama3-70B-Chinese-Chat, one of the 1st Llama3 70B models fine-tuned specifically for Chinese! 🏆 Llama3-70B-Chinese-Chat excels on C-Eval and CMMLU, surpassing ChatGPT and matching GPT-4. 🔥 Don't miss out on our model! Learn more: 🧵⬇️

shenzhi-wang/Mistral-7B-v0.3-Chinese-Chat · Hugging Face

4

25

98

Shenzhi Wang🌟

@ShenzhiWang_THU

5 months

🔥Introducing Mistral-7B-v0.3-Chinese-Chat, the 1st Mistral-7B-v0.3 model finetuned for Chinese&English. 🚀Our model consistently produces much better responses for Chinese prompts than the Mistral-7B-v0.3, and excels in math, roleplay, tool use, etc.

shenzhi-wang/Llama3-8B-Chinese-Chat-GGUF-f16 · Hugging Face

4

15

61

Shenzhi Wang🌟

@ShenzhiWang_THU

6 months

🎉 Our Llama3-8B-Chinese-Chat ranks 14th in overall trending on Huggingface, 2nd among Llama3 derivative models, and 1st among Llama3 Chinese derivative models! Thanks for your support! 🌟 I summarized our updates today in Fig. 3. See the details in the comments below:

7

2

17

Shenzhi Wang🌟

@ShenzhiWang_THU

6 months

🔥🔥🔥Update: We provide the official 8bit-quantized and fp16 versions of Llama3-8B-Chinese-Chat in the following links, respectively. You are welcome to have a try! 8bit-quantized: fp16:

ORPO: Monolithic Preference Optimization without Reference Model

Shenzhi Wang🌟

@ShenzhiWang_THU

6 months

🔥Introducing Llama3-8B-Chinese-Chat, the 1st llama3 model finetuned on English-Chinese datasets via ORPO. 🚀Our model consistently produces better responses for Chinese prompts than the Llama-3-8B-Insturct, and excels in logic, coding, math, and writing.

23

90

377

4

2

16

Shenzhi Wang🌟

@ShenzhiWang_THU

4 months

The fine-tuning algorithm we used is ORPO.

While recent preference alignment algorithms for language models have demonstrated promising results, supervised fine-tuning (SFT) remains imperative for achieving successful convergence. In this...

wangshenzhi/llama3-8b-chinese-chat-ollama-q4

1

0

11

Shenzhi Wang🌟

@ShenzhiWang_THU

5 months

🔥🔥Checkout DiveR-CT, Diversity-enhanced Red Teaming with Relaxing Constraints Highlights: 🚩New optimization by constrained RL 🚩Marked superiority in diversity and ASR 🚩Allowing dynamic control of objective weights 🚩Enhancing resiliency in blue team

Andrew Zhao

@AndrewZ45732491

5 months

🚩DiveR-CT, our latest automated red teaming method, forgoes reward maximization bias by constrained RL, also enhance semantic rewards to dynamically adapt. It delivers superior diversity, mitigates overoptimization and generates better data for safetuning

0

1

4

0

7

Shenzhi Wang🌟

@ShenzhiWang_THU

6 months

The Ollama model for the q4_0 GGUF version of Llama3-8B-Chinese-Chat-v2.1:

The ollama model for the 4bit-quantized GGUF version of llama3-8b-chinese-chat (https://huggingface.co/shenzhi-wang/Llama3-8B-Chinese-Chat-GGUF-4bit).

shenzhi-wang/Gemma-2-27B-Chinese-Chat at main

1

2

8

Shenzhi Wang🌟

@ShenzhiWang_THU

4 months

Various GGUF models for Gemma-2-27B-Chinese-Chat:

ORPO: Monolithic Preference Optimization without Reference Model

2

1

6

Shenzhi Wang🌟

@ShenzhiWang_THU

4 months

The fine-tuning algorithm we used is ORPO.

While recent preference alignment algorithms for language models have demonstrated promising results, supervised fine-tuning (SFT) remains imperative for achieving successful convergence. In this...

wangshenzhi/gemma2-27b-chinese-chat

1

0

5

Shenzhi Wang🌟

@ShenzhiWang_THU

6 months

Update: We provide the official 8bit-quantized and fp16 versions of Llama3-8B-Chinese-Chat in the following links, respectively. You are welcome to have a try! 8bit-quantized: fp16:

0

2

5

Shenzhi Wang🌟

@ShenzhiWang_THU

5 months

🚀 We've just launched the Llama3-70B-Chinese-Chat demo on Gitee! Big thanks to Gitee for their support! 😉 Feel free to give it a try! 🔗⬇️

Shenzhi Wang🌟

@ShenzhiWang_THU

5 months

🚀 Presenting Llama3-70B-Chinese-Chat, one of the 1st Llama3 70B models fine-tuned specifically for Chinese! 🏆 Llama3-70B-Chinese-Chat excels on C-Eval and CMMLU, surpassing ChatGPT and matching GPT-4. 🔥 Don't miss out on our model! Learn more: 🧵⬇️

4

25

98

1

2

4

Shenzhi Wang🌟

@ShenzhiWang_THU

4 months

The ollama model for Gemma-2-27B-Chinese-Chat:

The official ollama model for Gemma-2-27B-Chinese-Chat (https://huggingface.co/shenzhi-wang/Gemma-2-27B-Chinese-Chat).

gguf_models/gemma_2_chinese_chat_q8_0.gguf · shenzhi-wang/Gemma-2-9B-Chinese-Chat at main

1

0

5

Shenzhi Wang🌟

@ShenzhiWang_THU

4 months

GGUF q8_0 model of Gemma-2-9B-Chinese-Chat:

wangshenzhi/llama3-8b-chinese-chat-ollama-q8

1

0

5

Shenzhi Wang🌟

@ShenzhiWang_THU

4 months

We have included various examples generated by shenzhi-wang/Gemma-2-9B-Chinese-Chat, including examples of role playing, function calling, math, RuoZhiBa (弱智吧), safety, writing, and coding, etc. Have a look on our huggingface repo😆

1

0

5

Shenzhi Wang🌟

@ShenzhiWang_THU

6 months

(2/6) 🌟 GGUF 8bit quantized version of the Ollama model: , quick use: ollama run wangshenzhi/llama3-8b-chinese-chat-ollama-q8

The ollama model for the 8bit-quantized GGUF version of llama3-8b-chinese-chat (https://huggingface.co/shenzhi-wang/Llama3-8B-Chinese-Chat-GGUF-8bit).

Avalon's Game of Thoughts: Battle Against Deception through...

2

1

5

Shenzhi Wang🌟

@ShenzhiWang_THU

6 months

@huanggou7 @dotey 可以理解为4bit，8bit，16bit量化的版本，bit数越小越快，但性能可能会有所下降

2

1

3

Shenzhi Wang🌟

@ShenzhiWang_THU

7 months

🚀Jailbroken responses could potentially be mitigated through recursive thinking, which encourages LLMs to reconsider their initial outputs. 😄This idea is implemented in our proposed Recursive Contemplation (ReCon) framework. 🔥Check details below！

Recent breakthroughs in large language models (LLMs) have brought remarkable success in the field of LLM-as-Agent. Nevertheless, a prevalent assumption is that the information processed by LLMs is...

gguf_models/gemma_2_chinese_chat_f16.gguf · shenzhi-wang/Gemma-2-9B-Chinese-Chat at main

Anthropic

@AnthropicAI

7 months

New Anthropic research paper: Many-shot jailbreaking. We study a long-context jailbreaking technique that is effective on most large language models, including those developed by Anthropic and many of our peers. Read our blog post and the paper here:

82

354

2K

1

0

3

Shenzhi Wang🌟

@ShenzhiWang_THU

4 months

GGUF f16 model of Gemma-2-9B-Chinese-Chat:

wangshenzhi/gemma2-9b-chinese-chat

1

0

4

Shenzhi Wang🌟

@ShenzhiWang_THU

4 months

@GoogleDeepMind 🔥We introduced the first fine-tuned Gemma-2-9B model for Chinese&English users! 😄Welcome to have a try!

Shenzhi Wang🌟

@ShenzhiWang_THU

4 months

🔥Introducing Gemma-2-9B-Chinese-Chat: the 1st Gemma-2 model tailored for Chinese&English users, fine-tuned on >100K preference pairs! 🚀Our model excels in Chinese prompts, and shows improved logic, coding, math, and writing skills. More info: 🧵⬇️

10

33

190

0

1

4

Shenzhi Wang🌟

@ShenzhiWang_THU

4 months

The ollama model for Gemma2-9B-Chinese-Chat:

The official ollama model for Gemma-2-9B-Chinese-Chat (https://huggingface.co/shenzhi-wang/Gemma-2-9B-Chinese-Chat).

shenzhi-wang/Llama3-70B-Chinese-Chat-GGUF-4bit · Hugging Face

0

4

Shenzhi Wang🌟

@ShenzhiWang_THU

5 months

The official q4_0 GGUF files for Llama3-70B-Chinese-Chat:

ORPO: Monolithic Preference Optimization without Reference Model

1

4

Shenzhi Wang🌟

@ShenzhiWang_THU

5 months

The finetuning algorithm used is ORPO.

While recent preference alignment algorithms for language models have demonstrated promising results, supervised fine-tuning (SFT) remains imperative for achieving successful convergence. In this...

shenzhi-wang/Llama3-8B-Chinese-Chat-GGUF-4bit · Hugging Face

2

1

4

Shenzhi Wang🌟

@ShenzhiWang_THU

6 months

Examples (1/5)

0

4

Shenzhi Wang🌟

@ShenzhiWang_THU

6 months

The q4_0 GGUF version of Llama3-8B-Chinese-Chat-v2.1:

shenzhi-wang/Llama3-8B-Chinese-Chat-GGUF-f16 · Hugging Face

1

0

3

Shenzhi Wang🌟

@ShenzhiWang_THU

6 months

@minchoi Thank you for your summary! There’s another example 😆

Shenzhi Wang🌟

@ShenzhiWang_THU

6 months

🔥Introducing Llama3-8B-Chinese-Chat, the 1st llama3 model finetuned on English-Chinese datasets via ORPO. 🚀Our model consistently produces better responses for Chinese prompts than the Llama-3-8B-Insturct, and excels in logic, coding, math, and writing.

23

90

377

0

3

Shenzhi Wang🌟

@ShenzhiWang_THU

5 months

Results on C-Eval and CMMLU:

1

0

3

Shenzhi Wang🌟

@ShenzhiWang_THU

6 months

Thanks a lot for @llamafactory_ai 's assistance during training!

0

3

Shenzhi Wang🌟

@ShenzhiWang_THU

6 months

The f16 GGUF version of Llama3-8B-Chinese-Chat-v2.1:

wangshenzhi/llama3-70b-chinese-chat-ollama-q4

1

0

3

Shenzhi Wang🌟

@ShenzhiWang_THU

5 months

The ollama model for the 4bit-quantized GGUF version of llama3-70b-chinese-chat:

The ollama model for the 4bit-quantized GGUF version of llama3-70b-chinese-chat (https://huggingface.co/shenzhi-wang/Llama3-70B-Chinese-Chat-GGUF-4bit).

Avalon's Game of Thoughts: Battle Against Deception through...

1

0

3

Shenzhi Wang🌟

@ShenzhiWang_THU

7 months

@AnthropicAI 🚀Jailbroken responses could potentially be mitigated through recursive thinking, which encourages LLMs to reconsider their initial outputs. This idea is implemented in our proposed Recursive Contemplation (ReCon) framework. 🔥Check the details below！

Recent breakthroughs in large language models (LLMs) have brought remarkable success in the field of LLM-as-Agent. Nevertheless, a prevalent assumption is that the information processed by LLMs is...

wangshenzhi/llama3-8b-chinese-chat-ollama-fp16

1

0

3

Shenzhi Wang🌟

@ShenzhiWang_THU

5 months

If you’re in China, you can download our model from Gitee:

0

3

Shenzhi Wang🌟

@ShenzhiWang_THU

6 months

The Ollama model for the f16 GGUF version of Llama3-8B-Chinese-Chat-v2.1:

The ollama model for the f16 GGUF version of llama3-8b-chinese-chat (https://huggingface.co/shenzhi-wang/Llama3-8B-Chinese-Chat-GGUF-f16).

GitHub - Shenzhi-Wang/Llama3-Chinese-Chat: This is the first Chinese chat model specifically...

1

0

3

Shenzhi Wang🌟

@ShenzhiWang_THU

6 months

(1/6)🌟 Summary on how to download, use, and reproduce the Llama3-Chinese-8B-Chat model, along with subsequent updates:

This is the first Chinese chat model specifically fine-tuned for Chinese through ORPO based on the Meta-Llama-3-8B-Instruct model. - Shenzhi-Wang/Llama3-Chinese-Chat

github.com

0

1

3

Shenzhi Wang🌟

@ShenzhiWang_THU

6 months

The Ollama model for the q8_0 GGUF version of Llama3-8B-Chinese-Chat-v2.1:

wangshenzhi/llama3-8b-chinese-chat-ollama-q8

The ollama model for the 8bit-quantized GGUF version of llama3-8b-chinese-chat (https://huggingface.co/shenzhi-wang/Llama3-8B-Chinese-Chat-GGUF-8bit).

shenzhi-wang/Mistral-7B-v0.3-Chinese-Chat-4bit · Hugging Face

1

0

3

Shenzhi Wang🌟

@ShenzhiWang_THU

6 months

@9hills 感谢建议！扩充到32K是一个很好的下一步方向，然后100K数据我们会在进一步清洗后开源

1

0

3

Shenzhi Wang🌟

@ShenzhiWang_THU

4 months

@ycjcl We haven't thoroughly tested our model on a wide range of benchmarks yet, but we plan to do so. However, as far as I know, no current Chinese benchmarks effectively reflect both the model's instruction-following ability in Chinese and its code-switching issues.

0

1

Shenzhi Wang🌟

@ShenzhiWang_THU

6 months

typo: “ranks 14th in overall trending on Huggingface” should be “ranks 13th in overall trending on Huggingface” 🤣

0

2

Shenzhi Wang🌟

@ShenzhiWang_THU

6 months

Examples (2/5)

0

2

Shenzhi Wang🌟

@ShenzhiWang_THU

4 months

We have included various examples generated by shenzhi-wang/Gemma-2-27B-Chinese-Chat, including examples of role playing, function calling, math, RuoZhiBa (弱智吧), safety, writing, and coding, etc. Have a look on our huggingface repo😆

1

0

2

Shenzhi Wang🌟

@ShenzhiWang_THU

5 months

The official q4_0 GGUF files:

wangshenzhi/llama3-70b-chinese-chat-ollama-q8

1

0

2

Shenzhi Wang🌟

@ShenzhiWang_THU

4 months

@GoogleDeepMind 🔥After Gemma-2-9B-Chinese-Chat, we further introduce the first fine-tuned Gemma-2-27B model for Chinese&English users! 📷Welcome to have a try!

Shenzhi Wang🌟

@ShenzhiWang_THU

4 months

🔥 After the 9B model, we present Gemma-2-27B-Chinese-Chat: the 1st Gemma2 27B model optimized for Chinese&English, finetuned on >100K preference pairs! 🚀 Our model excels in Chinese, with improved logic, coding, math, and writing skills. More info:🧵⬇️

6

22

121

0

2

Shenzhi Wang🌟

@ShenzhiWang_THU

5 months

The examples for math:

1

0

2

Shenzhi Wang🌟

@ShenzhiWang_THU

6 months

🚀Our model ranked 7th overall on the Huggingface Trending Model leaderboard, 1st on the Huggingface Trending Chinese Model leaderboard, and 1st on the Huggignface Trending ORPO Model leaderboard! Thank you for your support! 😆Check out our v2.1 if you enjoy v1!

0

2

Shenzhi Wang🌟

@ShenzhiWang_THU

5 months

The ollama model for the 8bit-quantized GGUF version of llama3-70b-chinese-chat:

The ollama model for the 8bit-quantized GGUF version of llama3-70b-chinese-chat.

wangshenzhi/gemma2-27b-chinese-chat

1

0

2

Shenzhi Wang🌟

@ShenzhiWang_THU

4 months

@taikylin 已经有ollama啦😆

The official ollama model for Gemma-2-27B-Chinese-Chat (https://huggingface.co/shenzhi-wang/Gemma-2-27B-Chinese-Chat).

shenzhi-wang/Llama3-8B-Chinese-Chat-GGUF-f16 · Hugging Face

1

0

2

Shenzhi Wang🌟

@ShenzhiWang_THU

6 months

@meili145 Thank you for your interest in our model! We provide the official 8bit-quantized and fp16 versions of Llama3-8B-Chinese-Chat in the following links, respectively. You are welcome to have a try! 8bit-quantized: fp16:

shenzhi-wang/Mistral-7B-v0.3-Chinese-Chat-8bit · Hugging Face

1

0

2

Shenzhi Wang🌟

@ShenzhiWang_THU

5 months

The official q8_0 GGUF files:

shenzhi-wang/Mistral-7B-v0.3-Chinese-Chat-f16 · Hugging Face

1

0

2

Shenzhi Wang🌟

@ShenzhiWang_THU

4 months

Our Gemma-2-9B-Chinese-Chat:

Shenzhi Wang🌟

@ShenzhiWang_THU

4 months

🔥Introducing Gemma-2-9B-Chinese-Chat: the 1st Gemma-2 model tailored for Chinese&English users, fine-tuned on >100K preference pairs! 🚀Our model excels in Chinese prompts, and shows improved logic, coding, math, and writing skills. More info: 🧵⬇️

10

33

190

1

0

2

Shenzhi Wang🌟

@ShenzhiWang_THU

5 months

The examples for coding:

0

1

Shenzhi Wang🌟

@ShenzhiWang_THU

5 months

The official f16 GGUF files:

shenzhi-wang/Llama3-8B-Chinese-Chat-GGUF-f16 · Hugging Face

1

0

1

Shenzhi Wang🌟

@ShenzhiWang_THU

4 months

@linshuai_3344 请问您是什么样的场景下指令遵循不行呢？模型的指令遵循能力和模型大小以及场景有很大关系。如果对于某条指令，原版的模型和我们的汉化版本的模型都不行的话，这个可能受限于模型本身的大小以及原版模型的数据；如果原版模型可以但我们汉化版不行的话，欢迎您告知我们是什么场景，我们可以做定向优化。

1

0

1

Shenzhi Wang🌟

@ShenzhiWang_THU

6 months

(5/6) 🌟 GGUF FP16 version of the model:

shenzhi-wang/Llama3-8B-Chinese-Chat-GGUF-8bit · Hugging Face

0

1

Shenzhi Wang🌟

@ShenzhiWang_THU

6 months

The q8_0 GGUF version of Llama3-8B-Chinese-Chat-v2.1:

Chat with Llama3-8B-Chinese - a Hugging Face Space by llamafactory

1

0

1

Shenzhi Wang🌟

@ShenzhiWang_THU

5 months

@jack77844923 ollama q4版本大概需要2张A100 80G，一张我不确定是不是可以

1

0

1

Shenzhi Wang🌟

@ShenzhiWang_THU

6 months

Give our online demo a try! 😆

wangshenzhi/llama3-8b-chinese-chat-ollama-fp16

1

0

1

Shenzhi Wang🌟

@ShenzhiWang_THU

5 months

The examples for ruozhiba (弱智吧):

1

0

1

Shenzhi Wang🌟

@ShenzhiWang_THU

6 months

(6/6) 🌟 If you are in China, download here:

0

1

Shenzhi Wang🌟

@ShenzhiWang_THU

5 months

The examples for too use:

1

0

1

Shenzhi Wang🌟

@ShenzhiWang_THU

6 months

Examples (4/5)

0

1

Shenzhi Wang🌟

@ShenzhiWang_THU

6 months

Examples (5/5)

0

1

Shenzhi Wang🌟

@ShenzhiWang_THU

5 months

The examples for safety:

1

0

1

Shenzhi Wang🌟

@ShenzhiWang_THU

4 months

We have further released Gemma-2-27B-Chinese-Chat! If you love our Gemma-2-9B-Chinese-Chat, don't miss out on our latest Gemma-2-27B-Chinese-Chat! 😆

Shenzhi Wang🌟

@ShenzhiWang_THU

4 months

🔥 After the 9B model, we present Gemma-2-27B-Chinese-Chat: the 1st Gemma2 27B model optimized for Chinese&English, finetuned on >100K preference pairs! 🚀 Our model excels in Chinese, with improved logic, coding, math, and writing skills. More info:🧵⬇️

6

22

121

0

1

Shenzhi Wang🌟

@ShenzhiWang_THU

6 months

(3/6) 🌟 GGUF FP16 version of the Ollama model: , quick use: ollama run wangshenzhi/llama3-8b-chinese-chat-ollama-fp16

The ollama model for the f16 GGUF version of llama3-8b-chinese-chat (https://huggingface.co/shenzhi-wang/Llama3-8B-Chinese-Chat-GGUF-f16).