Fabian David Schmidt

Member of Technical Staff @ Cohere | PhD from Uni of Würzburg on multilinguality & multimodality | prev. Mila & LTL@UniCambridge

[Cohere @cohere]

Embed 5 and Rerank 5 are now live on OpenRouter!

Build faster, smarter retrieval applications with our state-of-the-art models through a single API. Two models, one bill, zero switching platforms.

Available in Pro and Fast tiers.

[Marius Mosbach @mariusmosbach]

I was recently interviewed by @CadeMetz for his latest piece on interpretability in the @nytimes 🤖🔍
You find it here 👉 https://t.co/AXkZFM25Sv

[Fabian David Schmidt @fdschmidt]

@MisterSnowCrash @cohere This should generally work well / as you intend with cutting off documents that score below 0.35. Above, documents typically are expected to have "useful information". This is lower than "answer" because for, e.g., multi-hop queries no document may contain the answer alone.

[Goran Glavaš @gg42554]

We, the WüNLP team (https://t.co/qCHab2wwm4), are heading to Budapest, where we will present six papers at EMNLP next week! If any of these spark your interest, just come talk to us :).

[Cohere @cohere]

Introducing Rerank 5

A few lines of code can change how you search. With this release, we’re advancing the frontier in AI retrieval and precision.

Available in Pro and Fast tiers today.

[Tiancheng Hu is looking for a job @tiancheng_hu]

Time flies and I am looking for full time research positions! I am finishing my PhD at Cambridge early 2027. My research focuses on simulating human behavior with LLMs, LLM personalization, evaluating and shaping model behavior, and understanding human–AI interaction. Please get in touch if you are hiring in relavant areas!

[Leon Engländer @LeonEnglaender]

I'm at COLM in SF this week! I work on agentic code gen at @cohere. If you want to talk about training code agents, DM me.

Our team is also hiring: https://t.co/OLpCt1hgkl

[Vaibhav Adlakha @ COLM 2026 @vaibhav_adlakha]

I will be at @COLM_conf this week, presenting LLM2Vec-Gen, come say hi!

📌 Poster session 2, Imperial Ballroom 4:30PM, poster #36

#COLM2026

[Cohere @cohere]

Introducing North 2.

Enterprises have been compromising on agentic AI for years: platforms that aren't scalable, models that aren't sovereign, control without enterprise-grade security.

That ends today, with 15+ new features in our biggest upgrade yet:

[Nils Reimers @Nils_Reimers]

Embedding model evaluations has many issues, so that sadly benchmarks like BEIR & MTEB became low signal.

With RCP-nDCG we are here to fix some of these:
https://t.co/p8FrnlktVA
https://t.co/9v46sZCN60

[Marius Mosbach @mariusmosbach]

I will be at #COLM2026 next week co-presenting:
1️⃣ https://t.co/MyThFxb90b (Tuesday)
2️⃣ https://t.co/uAevs8U6RB (Tuesday)
3️⃣ https://t.co/5zmD3Wa7MH (Thursday)

I will also be at the Actionable Interpretability workshop on Friday: https://t.co/1URwn4J3hf

Let's catch up ☕️ 🍻

[Cohere @cohere]

You might’ve noticed that our benchmarks for Embed 5 look a bit different 👀

Embed 5 is the first model family evaluated with RCP-nDCG@10, our latest methodology that improves on assessing real-world retrieval quality.

[Desmond Elliott @delliott]

📣 I'm hiring a Ph.D student through ELLIS to join my group at the University of Copenhagen. We work on tokenization-free language models (think: bytes, characters, and pixels). Your office will be at the Østervold Observatory in the beautiful Copenhagen Botanical Gardens.

[Fabian David Schmidt @fdschmidt]

@DonaldLamp84do @n0riskn0r3ward @benwtrent @cohere "Two annotators of the same document grade it within one step of each other in 77.8% of cases." More details in the paper https://t.co/Fn6EAHB45F and if you'd want to play with the data: https://t.co/gXaFJ2WS0t

[Benjamin Trent @benwtrent]

I really like @cohere 's RCP-nDCG idea. It's always bugged me how many gaps are in human labeled datasets (even the old standards). Using an LLM to fill in the gaps is very natural: https://t.co/yF68bugWow

[Cohere @cohere]

Introducing Cohere Embed 5: our new state-of-the-art family of embeddings models.

Get frontier capabilities with Embed 5 Pro or low-latency performance with Embed 5 Fast.

[Andreea Iana @iana_andreea]

Whose news, in which language, chosen how?

My interview on @DataSkeptic is now out. We got into how news recommenders shape what we read, why they work far better in English than in other languages, and what fairer information access could look like.

https://t.co/4ASid9CD5h

[Goran Glavaš @gg42554]

We're hiring! Three PhD/postdoc positions (different projects, topics, and constellations): (i) post-training of MoE LLMs, (ii) sample-efficient LM adaptation for low-resource domains, and (iii) VLA models for robotic manipulation.

More details at:
https://t.co/CMwuAOptX2

[Marius Mosbach @mariusmosbach]

I will be at #ICML2026 🇰🇷 presenting three papers on interpretability and model adaptation. Feel free to reach out if you want to chat! 🙂

P.S. I'm looking for motivated students to join my new group at Saarland University 🇩🇪: https://t.co/WJUfRpWBnQ

[Siva Reddy @sivareddyg]

Marius is a fantastic advisor, collaborator, colleague, and friend. Bittersweet to see him leave Montreal, but excited for all the amazing things he'll do at Saarland University and DFKI 🚀. He's hiring PhD students in LLM adaptation, interpretability, and continual learning. Apply now!