-
# Todo
+ [ ] Re-watch the AMI Toileting video and make notes:
https://montessori-ami.hosted.panopto.com/Panopto/Pages/Viewer.aspx?id=10bdc243-d568-4fb9-9525-af1c0080b5c6
- [WTF Drafts Index](https://davidblue.wtf/drafts/8C4596E3-0397-468C-A579-CF56CB7CFE06.htm…
-
> Especially towards the end of the second arc, discourse around growing up takes a a central role as a driving theme. (😏) And then suddenly we get the third season and everyone is in cars and their s…
-
### Issue URL (Ads)
[https://xmateur.com/embed/2509](https://adguardteam.github.io/AnonymousRedirect/redirect.html?url=https%3A%2F%2Fxmateur.com%2Fembed%2F2509)
### Comment
> https://xmateur.com…
-
https://mp.weixin.qq.com/s/PvbSXFrtcmI2G58kGNrNXw
-
### Reminder
- [X] I have read the README and searched the existing issues.
### Reproduction
deepspeed --hostfile=./hostfile --master_port=9901 src/train_bash.py \
--deepspeed ./ds_config.json…
-
Comment below with a well-developed question or comment about the reading for this week's workshop. These are individual questions and comments.
Please post your question by Wednesday 11:59 PM, and…
-
The README says
> This version is not compatible with the HF transformer implementation and must be used with SGLang or LLaVA implementation.
I hope to load and run the model with https://github…
zqzqz updated
4 months ago
-
我试了一下，在llama3-8b上使用magpie-pro-300k训练，在benchmark上差llama3-8b-it差很多，不是一星半点的差，在10个benchmark，mmlu gsm8k humaneval 等等上，能掉十几分到几十分不等，比如gsm8k，official instruct version 79.6 while magpie tuned version 40+ even …