6232 位用户此时在线
24小时点击排行 Top 10：
- 本站自动实时分享网络热点
- 24小时实时更新
- 所有言论不代表本站态度
- 欢迎对信息踊跃评论评分
- 评分越高，信息越新，排列越靠前
2
1
0
🤔 China's LLM open source game is splitting into two clear camps.
- Big tech (Alibaba, Tencent, ByteDance, Baidu) are going wide with foundation models across every parameter size.
- Startups (Zhipu, MiniMax, Moonshot) are going narrow, betting everything on code generation.
时政
(
twitter.com
)
由
Poe Zhao
提交
1
2
1
1
4
2
1
1
5
2
1
1
7
2
1
1
8
2
1
1
10
2
1
1
18
A new method was able to delete 40% of LLM layers with no drop in accuracy.
2
1
1
A new method was able to delete 40% of LLM layers with no drop in accuracy.
This makes them mich cheaper and faster.
The method combines pruning, quantization and PEFT.
They tested this across various open source models.
Each family of models had a maximum amount of layers…
时政
(
twitter.com)
•
Lior⚡
19
Reports of this incident went viral on 🇨🇳 social media a few days ago… During a demo of Huawei’s new LLM for text-to-image generation (generative AI) at the company’s Kunpeng Ascend Developer Conference on May 10, the presenter reportedly hit CTRL-C and some Python code was
3
2
2
Reports of this incident went viral on 🇨🇳 social media a few days ago… During a demo of Huawei’s new LLM for text-to-image generation (generative AI) at the company’s Kunpeng Ascend Developer Conference on May 10, the presenter reportedly hit CTRL-C and some Python code was
时政
(
twitter.com)
00:00:31
20
2
1
1
21
1
0
0
23
2
1
1
24
2
1
1