Patrick Lewis
@PSH_Lewis
Followers
5K
Following
2K
Media
21
Statuses
349
London-based AI/NLP Research Scientist. I lead RAG, Tool-use and Agents at Cohere. Previous Fundamental AI Research at Meta AI, FAIR, UCL AI
London, England
Joined August 2018
RT @_kire_kara_: šPsst! Want more faithful, verifiable and robust #LLM reasoning than with CoT, but using external solvers is meh? Our FLARā¦.
0
10
0
š
.
So happy to see @sarahookr and @PSH_Lewis on the AI 100 :) they have both contributed so much to @cohere and the field in general. Working with such impactful and ingenious people is one of the best perks of being at cohere :)
13
1
119
RT @s_hofstaetter: @karpathy @Ouponatime38 We build this at cohere š - claim level citations throughout our whole annotation, training, andā¦.
0
6
0
RT @osanseviero: @Teknium1 Command R+ is great for tool usage. We added tools (web search, url fetching, document parsing, image generationā¦.
0
2
0
RT @mervenoyann: document AI landed to @huggingface chat, powered by open-source models (@cohere Command-R+) with tool calling š ļøāØ. exploreā¦.
0
24
0
RT @seraphinagt: New paper! RAG-timely: *MultiContrievers: Analysis of Dense Retrieval Representations* w/ @PSH_Lewis @EntilZhaPR @JaneDwivā¦.
arxiv.org
Dense retrievers compress source documents into (possibly lossy) vector representations, yet there is little analysis of what information is lost versus preserved, and how it affects downstream...
0
14
0
RT @natolambert: Evaluation in AI has only works when a problem space is sufficiently scoped. We no longer evaluate LLMs on specific thingā¦.
0
6
0
RT @soldni: maybe controversial, but:. Prompts are a property of models, not benchmarks. Assuming the opposite leads to benchmark hacking.ā¦.
0
4
0
New paper from our team, led by @pat_verga .Are you:.* Doing evaluation with LLMs? .* Using a huge model?.* Worried about self-recognition?. Try an ensemble of smaller LLMs. Use a PoLL: less biased, faster, 7x cheaper. Works great on QA & Arena-hard evals.
9
38
184
RT @victormustar: Command R+ is underrated for everyday use š. The speed & the balance of good reasoning, concise answers, and nice writinā¦.
0
11
0
RT @_valsai: Just updated our benchmark to include the new Cohere Command models and DBRX! See results at
0
2
0
RT @aidangomez: Introducing Rerank 3! Our latest model focused on powering much more complex and accurate search. It's the fastest, cheapeā¦.
0
120
0
RT @NandoDF: Iām a huge fan of LLM efforts that emphasise multilingualism because they make AI far more inclusive and accessible. Congratulā¦.
0
12
0
RT @LangChainAI: š @cohere Integration Package š¦. Hot off the heels of Cohere's v5 sdk supporting tool calling and their new Command R+ modā¦.
0
53
0
RT @W4ngatang: This drove me crazy for a while: We had internal experiments showing RM > LLM for evaluation, which felt really counterintuiā¦.
0
7
0
RT @jordnb: Cohere's Chat interface with new āR+ model looks. really good!. š search + cite.š» web browsing.š python . Watch this. https:/ā¦.
0
17
0
RT @itsSandraKublik: mind-blow by how good ā R+ multi-step tool use is š¤Æ. rewrites my sloppy query -> fetches numbers -> plots them with ciā¦.
0
25
0