Anthony Susevski
@asusevski
Followers
365
Following
6K
Media
148
Statuses
2K
ml enjoyer. find it from within or be without
Joined March 2021
RT @vitransformer: New blog post: We've never enjoyed working on Kernels more than this. We have some very fast AI-generated kernels with….
0
55
0
RT @finbarrtimbers: horrifying bug of the day is finding out the vllm and huggingface produce significantly different logprobs. https://t.c….
0
41
0
RT @ShashwatGoel7: There's been a hole at the heart of #LLM evals, and we can now fix it. 📜New paper: Answer Matching Outperforms Multiple….
0
30
0
RT @rise24546323: @osanseviero We have now traced down the main issue of poor quality in Gemma 3n MobileNet V5 to an incorrect conv layers….
0
4
0
RT @jjrichardtang: Hiring Waterloo 6 interns to join @rootlyhq here in Toronto. We are building an AI-native incident management platform….
0
9
0
RT @giffmana: Or, in other words, Gemini2.5 Pro succeeds at 30% of real world office tasks. That's pretty good, considering this is the wo….
0
29
0
RT @CherylolGuo: ❤️🌎 Introducing CARE: Multilingual Multicultural Human Preference Learning.3490 culturally relevant prompts + 31.7k Human/….
0
13
0
RT @tech_optimist: Great summary of why SWE-bench is flawed, and more generally, why all benchmarks are flawed. Always build your own eval….
0
1
0
Who’s building a CLI tool that re-directs stderr to Claude to classify the error as skill issue or not.
1
0
2
RT @kennylpeng: Are LLMs correlated when they make mistakes? In our new ICML paper, we answer this question using responses of >350 LLMs. W….
0
33
0
RT @_WEEXIAO: It’s easy to fine-tune small models w/ RL to outperform foundation models on vertical tasks. We’re open sourcing Osmosis-App….
0
123
0
attenzione interns (if you're cracked listen up):. @rootlyhq, who has SO GRACIOUSLY provided funding for Iced Coffees at Papers in the park this Saturday, is hiring. DM @jjrichardtang to apply.
2
2
9
RT @swyx: icymi, openai also open sourced how their deep research prompt rewriter works + the full prompts today. you can now build your ow….
0
171
0
RT @lusxvr: Today, we are open-sourcing our pipeline to deduplicate large-scale image datasets. On one GPU, we can deduplicate 10k images….
0
99
0
