Strategic Navigation or Stochastic Search? How Agents and Humans Reason Over Document Collections
Paper
• 2603.12180 • Published
• 39
AI & ML interests
None defined yet.
Recent Activity
View all activity
clefourrier
authored a paper about 12 hours ago
Post
3733
👉 What happened in AI in 2025? 👈
We prepared the 2025 version of the HF AI Timeline Grid, highlighting open vs API-based model releases, and allowing you to browse and filter by access, modality, and release type!
Play with it here:
2025-ai-timeline/2025-ai-timeline
Here's my personal quarterly TL;DR:
1️⃣ Q1 — Learning to Reason
Deepseek not only releases a top-notch reasoning model, but shows how to train them and compete with closed frontier models. OpenAI debuts Deep Research.
Significant milestones: DeepSeek R1 & R1-Zero, Qwen 2.5 VL, OpenAI Deep Research, Gemini 2.5 Pro (experimental)
2️⃣ Q2 — Multimodality and Coding
More LLMs embrace multimodality by default, and there's a surge in coding agents. Strong vision, audio, and generative models emerge.
Significant milestones: Llama 4, Qwen 3, Imagen 4, OpenAI Codex, Google Jules, Claude 4
3️⃣ Q3 — "Gold" rush, OpenAI opens up, the community goes bananas
Flagship models get gold in Math olympiads and hard benchmarks. OpenAI releases strong open source models and Google releases the much anticipated nano-banana for image generation and editing. Agentic workflows become commonplace.
Significant milestones: Gemini and OpenAI IMO Gold, gpt-oss, Gemini 2.5 Flash Image, Grok 4, Claude Sonnet 4.5
4️⃣ Q4 — Mistral returns, leaderboard hill-climbing
Mistral is back with updated model families. All labs release impressive models to wrap up the year!
Significant milestones: Claude Opus 4.5, DeepSeek Math V2, FLUX 2, GPT 5.1, Kimi K2 Thinking, Nano Banana Pro, GLM 4.7, Gemini 3, Mistral 3, MiniMax M2.1 🤯
Credits
🙏 NHLOCAL for the source data https://github.com/NHLOCAL/AiTimeline
🫡 @reach-vb for the original idea, design and recipe
🙌 @ariG23498 and yours truly for compiling and verifying the 2025 edition
🥳 Here's to 2026, wishing it becomes the best year ever for open releases and on-device-first use-cases! 🥂
We prepared the 2025 version of the HF AI Timeline Grid, highlighting open vs API-based model releases, and allowing you to browse and filter by access, modality, and release type!
Play with it here:
2025-ai-timeline/2025-ai-timeline
Here's my personal quarterly TL;DR:
1️⃣ Q1 — Learning to Reason
Deepseek not only releases a top-notch reasoning model, but shows how to train them and compete with closed frontier models. OpenAI debuts Deep Research.
Significant milestones: DeepSeek R1 & R1-Zero, Qwen 2.5 VL, OpenAI Deep Research, Gemini 2.5 Pro (experimental)
2️⃣ Q2 — Multimodality and Coding
More LLMs embrace multimodality by default, and there's a surge in coding agents. Strong vision, audio, and generative models emerge.
Significant milestones: Llama 4, Qwen 3, Imagen 4, OpenAI Codex, Google Jules, Claude 4
3️⃣ Q3 — "Gold" rush, OpenAI opens up, the community goes bananas
Flagship models get gold in Math olympiads and hard benchmarks. OpenAI releases strong open source models and Google releases the much anticipated nano-banana for image generation and editing. Agentic workflows become commonplace.
Significant milestones: Gemini and OpenAI IMO Gold, gpt-oss, Gemini 2.5 Flash Image, Grok 4, Claude Sonnet 4.5
4️⃣ Q4 — Mistral returns, leaderboard hill-climbing
Mistral is back with updated model families. All labs release impressive models to wrap up the year!
Significant milestones: Claude Opus 4.5, DeepSeek Math V2, FLUX 2, GPT 5.1, Kimi K2 Thinking, Nano Banana Pro, GLM 4.7, Gemini 3, Mistral 3, MiniMax M2.1 🤯
Credits
🙏 NHLOCAL for the source data https://github.com/NHLOCAL/AiTimeline
🫡 @reach-vb for the original idea, design and recipe
🙌 @ariG23498 and yours truly for compiling and verifying the 2025 edition
🥳 Here's to 2026, wishing it becomes the best year ever for open releases and on-device-first use-cases! 🥂
- 2 replies
hfkwr
authored a paper 5 months ago
Post
8163
We're kick-starting the process of Transformers v5, with @ArthurZ and @cyrilvallez !
v5 should be significant: we're using it as a milestone for performance optimizations, saner defaults, and a much cleaner code base worthy of 2025.
Fun fact: v4.0.0-rc-1 came out on Nov 19, 2020, nearly five years ago!
v5 should be significant: we're using it as a milestone for performance optimizations, saner defaults, and a much cleaner code base worthy of 2025.
Fun fact: v4.0.0-rc-1 came out on Nov 19, 2020, nearly five years ago!
- 6 replies
clefourrier
posted an
update 10 months ago
Post
2410
Always surprised that so few people actually read the FineTasks blog, on
✨how to select training evals with the highest signal✨
If you're serious about training models without wasting compute on shitty runs, you absolutely should read it!!
An high signal eval actually tells you precisely, during training, how wel & what your model is learning, allowing you to discard the bad runs/bad samplings/...!
The blog covers in depth prompt choice, metrics, dataset, across languages/capabilities, and my fave section is "which properties should evals have"👌
(to know on your use case how to select the best evals for you)
Blog: HuggingFaceFW/blogpost-fine-tasks
✨how to select training evals with the highest signal✨
If you're serious about training models without wasting compute on shitty runs, you absolutely should read it!!
An high signal eval actually tells you precisely, during training, how wel & what your model is learning, allowing you to discard the bad runs/bad samplings/...!
The blog covers in depth prompt choice, metrics, dataset, across languages/capabilities, and my fave section is "which properties should evals have"👌
(to know on your use case how to select the best evals for you)
Blog: HuggingFaceFW/blogpost-fine-tasks
-
- 2 replies
pcuenq
authored a paper 11 months ago
clefourrier
posted an
update about 1 year ago
Post
2708
Gemma3 family is out! Reading the tech report, and this section was really interesting to me from a methods/scientific fairness pov.
Instead of doing over-hyped comparisons, they clearly state that **results are reported in a setup which is advantageous to their models**.
(Which everybody does, but people usually don't say)
For a tech report, it makes a lot of sense to report model performance when used optimally!
On leaderboards on the other hand, comparison will be apples to apples, but in a potentially unoptimal way for a given model family (like some user interact sub-optimally with models)
Also contains a cool section (6) on training data memorization rate too! Important to see if your model will output the training data it has seen as such: always an issue for privacy/copyright/... but also very much for evaluation!
Because if your model knows its evals by heart, you're not testing for generalization.
Instead of doing over-hyped comparisons, they clearly state that **results are reported in a setup which is advantageous to their models**.
(Which everybody does, but people usually don't say)
For a tech report, it makes a lot of sense to report model performance when used optimally!
On leaderboards on the other hand, comparison will be apples to apples, but in a potentially unoptimal way for a given model family (like some user interact sub-optimally with models)
Also contains a cool section (6) on training data memorization rate too! Important to see if your model will output the training data it has seen as such: always an issue for privacy/copyright/... but also very much for evaluation!
Because if your model knows its evals by heart, you're not testing for generalization.
Post
8377
SmolVLM-2 and SigLIP-2 are now part of
They're added on top of the v4.49.0 release, and can be installed from the following tags:
This marks a new beginning for the release process of transformers. For the past five years, we've been doing monthly releases featuring many models (v4.49.0, the latest release, features 9 new architectures).
Starting with SmolVLM-2 & SigLIP2, we'll now additionally release tags supporting new models on a stable branch. These models are therefore directly available for use by installing from the tag itself. These tags will continue to be updated with fixes applied to these models.
Going forward, continue expecting software releases following semantic versioning: v4.50.0 will have ~10 new architectures compared to v4.49.0, as well as a myriad of new features, improvements and bug fixes. Accompanying these software releases, we'll release tags offering brand new models as fast as possible, to make them accessible to all immediately.
transformers in dedicated releases!They're added on top of the v4.49.0 release, and can be installed from the following tags:
v4.49.0-SmolVLM-2 and v4.49.0-SigLIP-2.This marks a new beginning for the release process of transformers. For the past five years, we've been doing monthly releases featuring many models (v4.49.0, the latest release, features 9 new architectures).
Starting with SmolVLM-2 & SigLIP2, we'll now additionally release tags supporting new models on a stable branch. These models are therefore directly available for use by installing from the tag itself. These tags will continue to be updated with fixes applied to these models.
Going forward, continue expecting software releases following semantic versioning: v4.50.0 will have ~10 new architectures compared to v4.49.0, as well as a myriad of new features, improvements and bug fixes. Accompanying these software releases, we'll release tags offering brand new models as fast as possible, to make them accessible to all immediately.
- 1 reply
clefourrier
authored a paper about 1 year ago
clefourrier
authored a paper over 1 year ago
wukaixingxp
updated a dataset over 1 year ago
SaylorTwift
posted an
update over 1 year ago
Post
1510
How do I test an LLM for my unique needs?
If you work in finance, law, or medicine, generic benchmarks are not enough.
This blog post uses Argilla, Distilllabel and 🌤️Lighteval to generate evaluation dataset and evaluate models.
https://github.com/argilla-io/argilla-cookbook/blob/main/domain-eval/README.md
If you work in finance, law, or medicine, generic benchmarks are not enough.
This blog post uses Argilla, Distilllabel and 🌤️Lighteval to generate evaluation dataset and evaluate models.
https://github.com/argilla-io/argilla-cookbook/blob/main/domain-eval/README.md
wukaixingxp
authored a paper over 1 year ago
clefourrier
authored 2 papers over 1 year ago
Post
10301
OpenELM in Core ML
Apple recently released a set of efficient LLMs in sizes varying between 270M and 3B parameters. Their quality, according to benchmarks, is similar to OLMo models of comparable size, but they required half the pre-training tokens because they use layer-wise scaling, where the number of attention heads increases in deeper layers.
I converted these models to Core ML, for use on Apple Silicon, using this script: https://gist.github.com/pcuenca/23cd08443460bc90854e2a6f0f575084. The converted models were uploaded to this community in the Hub for anyone that wants to integrate inside their apps: corenet-community/openelm-core-ml-6630c6b19268a5d878cfd194
The conversion was done with the following parameters:
- Precision: float32.
- Sequence length: fixed to 128.
With swift-transformers (https://github.com/huggingface/swift-transformers), I'm getting about 56 tok/s with the 270M on my M1 Max, and 6.5 with the largest 3B model. These speeds could be improved by converting to
I'm also looking at optimizing inference using an experimental kv cache in swift-transformers. It's a bit tricky because the layers have varying number of attention heads, but I'm curious to see how much this feature can accelerate performance in this model family :)
Regarding the instruct fine-tuned models, I don't know the chat template that was used. The models use the Llama 2 tokenizer, but the Llama 2 chat template, or the default Alignment Handbook one that was used to train, are not recognized. Any ideas on this welcome!
Apple recently released a set of efficient LLMs in sizes varying between 270M and 3B parameters. Their quality, according to benchmarks, is similar to OLMo models of comparable size, but they required half the pre-training tokens because they use layer-wise scaling, where the number of attention heads increases in deeper layers.
I converted these models to Core ML, for use on Apple Silicon, using this script: https://gist.github.com/pcuenca/23cd08443460bc90854e2a6f0f575084. The converted models were uploaded to this community in the Hub for anyone that wants to integrate inside their apps: corenet-community/openelm-core-ml-6630c6b19268a5d878cfd194
The conversion was done with the following parameters:
- Precision: float32.
- Sequence length: fixed to 128.
With swift-transformers (https://github.com/huggingface/swift-transformers), I'm getting about 56 tok/s with the 270M on my M1 Max, and 6.5 with the largest 3B model. These speeds could be improved by converting to
float16. However, there's some precision loss somewhere and generation doesn't work in float16 mode yet. I'm looking into this and will keep you posted! Or take a look at this issue if you'd like to help: https://github.com/huggingface/swift-transformers/issues/95I'm also looking at optimizing inference using an experimental kv cache in swift-transformers. It's a bit tricky because the layers have varying number of attention heads, but I'm curious to see how much this feature can accelerate performance in this model family :)
Regarding the instruct fine-tuned models, I don't know the chat template that was used. The models use the Llama 2 tokenizer, but the Llama 2 chat template, or the default Alignment Handbook one that was used to train, are not recognized. Any ideas on this welcome!
-
- 5 replies
clefourrier
posted an
update almost 2 years ago
Post
6184
In a basic chatbots, errors are annoyances. In medical LLMs, errors can have life-threatening consequences 🩸
It's therefore vital to benchmark/follow advances in medical LLMs before even thinking about deployment.
This is why a small research team introduced a medical LLM leaderboard, to get reproducible and comparable results between LLMs, and allow everyone to follow advances in the field.
openlifescienceai/open_medical_llm_leaderboard
Congrats to @aaditya and @pminervini !
Learn more in the blog: https://huggingface.co/blog/leaderboard-medicalllm
It's therefore vital to benchmark/follow advances in medical LLMs before even thinking about deployment.
This is why a small research team introduced a medical LLM leaderboard, to get reproducible and comparable results between LLMs, and allow everyone to follow advances in the field.
openlifescienceai/open_medical_llm_leaderboard
Congrats to @aaditya and @pminervini !
Learn more in the blog: https://huggingface.co/blog/leaderboard-medicalllm
clefourrier
posted an
update almost 2 years ago
Post
4800
Contamination free code evaluations with LiveCodeBench! 🖥️
LiveCodeBench is a new leaderboard, which contains:
- complete code evaluations (on code generation, self repair, code execution, tests)
- my favorite feature: problem selection by publication date 📅
This feature means that you can get model scores averaged only on new problems out of the training data. This means... contamination free code evals! 🚀
Check it out!
Blog: https://huggingface.co/blog/leaderboard-livecodebench
Leaderboard: livecodebench/leaderboard
Congrats to @StringChaos @minimario @xu3kev @kingh0730 and @FanjiaYan for the super cool leaderboard!
LiveCodeBench is a new leaderboard, which contains:
- complete code evaluations (on code generation, self repair, code execution, tests)
- my favorite feature: problem selection by publication date 📅
This feature means that you can get model scores averaged only on new problems out of the training data. This means... contamination free code evals! 🚀
Check it out!
Blog: https://huggingface.co/blog/leaderboard-livecodebench
Leaderboard: livecodebench/leaderboard
Congrats to @StringChaos @minimario @xu3kev @kingh0730 and @FanjiaYan for the super cool leaderboard!
clefourrier
posted an
update almost 2 years ago
Post
2281
🆕 Evaluate your RL agents - who's best at Atari?🏆
The new RL leaderboard evaluates agents in 87 possible environments (from Atari 🎮 to motion control simulations🚶and more)!
When you submit your model, it's run and evaluated in real time - and the leaderboard displays small videos of the best model's run, which is super fun to watch! ✨
Kudos to @qgallouedec for creating and maintaining the leaderboard!
Let's find out which agent is the best at games! 🚀
open-rl-leaderboard/leaderboard
The new RL leaderboard evaluates agents in 87 possible environments (from Atari 🎮 to motion control simulations🚶and more)!
When you submit your model, it's run and evaluated in real time - and the leaderboard displays small videos of the best model's run, which is super fun to watch! ✨
Kudos to @qgallouedec for creating and maintaining the leaderboard!
Let's find out which agent is the best at games! 🚀
open-rl-leaderboard/leaderboard
clefourrier
posted an
update almost 2 years ago
Post
2266
Fun fact about evaluation, part 2!
How much do scores change depending on prompt format choice?
Using different prompts (all present in the literature, from
10 points for a single model!
Keep in mind that we only changed the prompt, not the evaluation subsets, etc.
Again, this confirms that evaluation results reported without their details are basically bullshit.
Prompt format on the x axis, all these evals look at the logprob of either "choice A/choice B..." or "A/B...".
Incidentally, it also changes model rankings - so a "best" model might only be best on one type of prompt...
How much do scores change depending on prompt format choice?
Using different prompts (all present in the literature, from
Prompt question? to Question: prompt question?\nChoices: enumeration of all choices\nAnswer: ), we get a score range of...10 points for a single model!
Keep in mind that we only changed the prompt, not the evaluation subsets, etc.
Again, this confirms that evaluation results reported without their details are basically bullshit.
Prompt format on the x axis, all these evals look at the logprob of either "choice A/choice B..." or "A/B...".
Incidentally, it also changes model rankings - so a "best" model might only be best on one type of prompt...