We want to THANK @lmsysorg for ur timely inclusion the day we launched Yi-Large (an unplanned clash w/ GPT-4o, ha). Also very encouraging to see your increasing inclusion of diversified languages (Chinese, French…) and advanced capabilities e.g. launching the latest Hard
RESEARCH
-
Prompt Engineering Evolution: From Art to Science for AI
By
–
it doesnt kill prompt engineering, but by god it simplifies tweaking model behavior from a vague art for wordcels toward a science for tensor rotators. more in ainews
-
Anthropic’s Progress Toward Productizable AI Interpretability
By
–
IMO @AnthropicAI is very close to making a breakthrough in productizable interpretability. For ~4 years all we've had to really control LLMs is temperature/top_p and logit bias. We recently got `seed` and constrained structured output, with `interactive=false` on the way. But
-
Anthropic’s interpretability team delivers impressive research work
By
–
Some really nice work by Anthropic's interpretability team. Kudos to @ch402 (former intern!) and everyone else involved!
-
Resources on AI assistants problems and research insights
By
–
You can read more in our article here: https://
catalystjournal.org/index.php/cata
lyst/article/view/29949/26063
…
And here's an important article from DeepMind on the problems with AI assistants: https://
drive.google.com/file/d/13mLAeN
TL7TbscYuGdEXTFHemWGd7HLhn/view
… -
LayerSkip: Optimizing LLM Inference with Speculative Decoding
By
–
LayerSkip is a new paper from @AIatMeta that improves LLM inference by merging speculative decoding and early exit. Authors @m_elhoushi and @AkshatS07 are on alphaXiv to answer questions!
-
GPT-5 to be a whale: OpenAI’s next model evolution
By
–
¡GPT-5 SERÁ UNA BALLENA! Por primera vez en un escenario se ha hecho mención al próximo modelo de OpenAI, usando una analogía marina para comparar las diferencias de computación dedicada: > GPT-3: Fue un tiburón
> GPT-4: Fue una orca
> GPT-5: Será una ballena! -
Anthropic Interpretability Paper on Scaling Monosemanticity
By
–
The new interpretability paper from Anthropic is totally based. Feels like analyzing an alien life form. If you only read one 90-min-read paper today, it has to be this one https://
transformer-circuits.pub/2024/scaling-m
onosemanticity/index.html
… -
GPT-5 Coming Soon: Sam Altman Takes the Stage
By
–
GPT-5 IS COMING! (soon) Se sube Sam Altman al escenario.
-
Programmatic AI Data Development Keynote at Generative AI Summit
By
–
Let's do this! @ajratner
's keynote speaks to the future of programmatic AI data development at London's premiere AI event #generativeaisummit!