Kuzman Ganchev's Avatar

Kuzman Ganchev

@ganchev.bsky.social

Research Scientist at GoogleDeepMind (formerly at Google Research). UPenn graduate.

1,441 Followers  |  33 Following  |  3 Posts  |  Joined: 25.10.2024  |  1.7666

Latest posts by ganchev.bsky.social on Bluesky

Post image Post image Post image

Study in Nature: β€œAcross 30 out of 32 evaluation axes from the specialist physician perspective & 25 out of 26 evaluation axes from the patient-actor perspective, AMIE [Google Medical LLM] was rated superior to PCPs [primary care docs] while being non-inferior on the rest.”

(& AIME is an older LLM)

04.05.2025 13:27 β€” πŸ‘ 70    πŸ” 15    πŸ’¬ 4    πŸ“Œ 7
Preview
Gemma explained: What’s new in Gemma 3- Google Developers Blog Google's Gemma 3 model includes vision-language support and architectural changes for resource-friendly multimodal language models.

Gemma 3 explained: Longer context, image support, and a new 1B model. β†’ goo.gle/4lV8iaw

Other key enhancements:
πŸ”Έ Best model that fits in a single consumer GPU or TPU host
πŸ”Έ KV-cache memory reduction with 5-to-1 interleaved attention
πŸ”Έ And more!

Read the blog for the full details on Gemma 3.

30.04.2025 21:46 β€” πŸ‘ 22    πŸ” 8    πŸ’¬ 1    πŸ“Œ 0

There's a link to a really nice interactive viewer for a sample of the data (will only make sense after you read the post). There's some examples that I would have expected (where something is implied but not directly stated) but also a surprising number of kind of topical things.

17.12.2024 16:12 β€” πŸ‘ 3    πŸ” 1    πŸ’¬ 0    πŸ“Œ 0

Want to get started using PaliGemma 2?

🎀 developers.googleblog.com/en/introduci...
πŸ€— huggingface.co/blog/paligem...
πŸ’Ύ kaggle.com/models/googl...
πŸ”§ github.com/google-resea...

7/7

05.12.2024 18:19 β€” πŸ‘ 7    πŸ” 1    πŸ’¬ 0    πŸ“Œ 0
Preview
GitHub - varungodbole/prompt-tuning-playbook: A playbook for effectively prompting post-trained LLMs A playbook for effectively prompting post-trained LLMs - varungodbole/prompt-tuning-playbook

Wanted to share that Varun Godbole recently released a prompting playbook. The title says prompt tuning, but this is text prompts, not soft prompts.

github.com/varungodbole...

11.11.2024 15:51 β€” πŸ‘ 14    πŸ” 7    πŸ’¬ 0    πŸ“Œ 0
Preview
ALTA: Compiler-Based Analysis of Transformers We propose a new programming language called ALTA and a compiler that can map ALTA programs to Transformer weights. ALTA is inspired by RASP, a language proposed by Weiss et al. (2021), and Tracr (Lin...

I’m pretty excited about this one!

ALTA is A Language for Transformer Analysis.

Because ALTA programs can be compiled to transformer weights, it provides constructive proofs of transformer expressivity. It also offers new analytic tools for *learnability*.

arxiv.org/abs/2410.18077

24.10.2024 03:31 β€” πŸ‘ 53    πŸ” 16    πŸ’¬ 2    πŸ“Œ 0
Preview
Zed - The editor for what's next Zed is a high-performance, multiplayer code editor from the creators of Atom and Tree-sitter.

Not news, but I recently saw the zed.dev demo and it looks amazing. Has anyone used it or something similar?

25.10.2024 14:43 β€” πŸ‘ 3    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

@ganchev is following 20 prominent accounts