Tash builds AI
Region: IE
Thursday 02 July 2026 13:03:32 GMT
5944
181
9
21
Music
Download
Comments
cc_pz :
I mean not sure if it was cheating or not but if it’s legit it’s cool
2026-07-02 20:07:02
2
Chamy :
Can you share it ?
2026-07-11 11:11:53
2
Matt 🇵🇷 :
I tried and compared it with a few other models, it was dumber. Bonsai looks promising
2026-07-23 12:25:19
0
Enrico Wolf39 :
It is actually seen in the words he used the qwen 3.6 35B mtp speculate decoding and connected it to the Ornith 35B that’s based on qwen 3.5. why gives that more? Because mtp is a wording predictor mode that guesses on the next words needed in response. I thought about it too but had headaches to get ornith running and sadly I used the 9B dense yet for testing and not happy, cause I had to use Q4 what’s really bad in tool use. Quantisation is best way said biggest crap u can use. Unless this issue isn’t overcome at least q8 is the minimum quantity usable no matter what model is used.
2026-07-02 20:06:38
1
Dr.Skills | AI Labs :
Pouvez-vous faire une vidéo plus en-depth pour expliquer ce que vous voulez dire par la chirurgie de l'œil ? Vous parlez en fait du fait qu'elle a été capable d'aller en-depth dans le modèle et de donner de nouvelles façons d'améliorer leurs capacités. Est-ce que vous
2026-07-11 14:18:57
0
888yea :
Can you explain better what you did?
2026-07-02 18:19:00
1
Wetime.ai :
https://github.com/corbenicai/aneirin
2026-07-02 20:21:04
0
Derrick Muyanzi| Ai and Tech :
how I wish I had my own c to do this cool stuff. good job bro
2026-07-02 18:22:42
2
To see more videos from user @tash.builds.ai, please go to the Tikwm
homepage.
