Article
Vessel Browser: The Open Source Browser Designed for Autonomous Agents
•
1
Tyler Williams PRO
unmodeled-tyler
AI & ML interests
AI research engineer & solo operator of VANTA Research/Quanta Intellect
Recent Activity
upvoted an article 38 minutes ago
Vessel Browser: The Open Source Browser Designed for Autonomous Agents reacted to theirpost with 🚀 about 2 hours ago
Just published a new blog!: https://huggingface.co/blog/unmodeled-tyler/vessel-browser-for-agents
I do a lightweight overview of the open source AI-native browser I've been building called Vessel Browser!
It recently passed 50 stars on Github, and I tremendously appreciate everyone that's given it a try so far! I'm in it for the long haul with Vessel Browser so expect many, many more improvements to come.
If you'd like to check out the repo: https://github.com/unmodeled-tyler/vessel-browser posted an update about 2 hours ago
Just published a new blog!: https://huggingface.co/blog/unmodeled-tyler/vessel-browser-for-agents
I do a lightweight overview of the open source AI-native browser I've been building called Vessel Browser!
It recently passed 50 stars on Github, and I tremendously appreciate everyone that's given it a try so far! I'm in it for the long haul with Vessel Browser so expect many, many more improvements to come.
If you'd like to check out the repo: https://github.com/unmodeled-tyler/vessel-browserOrganizations
upvoted an article 38 minutes ago
reacted to their post with 🚀 about 2 hours ago
Post
6
Just published a new blog!: https://huggingface.co/blog/unmodeled-tyler/vessel-browser-for-agents
I do a lightweight overview of the open source AI-native browser I've been building called Vessel Browser!
It recently passed 50 stars on Github, and I tremendously appreciate everyone that's given it a try so far! I'm in it for the long haul with Vessel Browser so expect many, many more improvements to come.
If you'd like to check out the repo: https://github.com/unmodeled-tyler/vessel-browser
I do a lightweight overview of the open source AI-native browser I've been building called Vessel Browser!
It recently passed 50 stars on Github, and I tremendously appreciate everyone that's given it a try so far! I'm in it for the long haul with Vessel Browser so expect many, many more improvements to come.
If you'd like to check out the repo: https://github.com/unmodeled-tyler/vessel-browser
posted an update about 2 hours ago
Post
6
Just published a new blog!: https://huggingface.co/blog/unmodeled-tyler/vessel-browser-for-agents
I do a lightweight overview of the open source AI-native browser I've been building called Vessel Browser!
It recently passed 50 stars on Github, and I tremendously appreciate everyone that's given it a try so far! I'm in it for the long haul with Vessel Browser so expect many, many more improvements to come.
If you'd like to check out the repo: https://github.com/unmodeled-tyler/vessel-browser
I do a lightweight overview of the open source AI-native browser I've been building called Vessel Browser!
It recently passed 50 stars on Github, and I tremendously appreciate everyone that's given it a try so far! I'm in it for the long haul with Vessel Browser so expect many, many more improvements to come.
If you'd like to check out the repo: https://github.com/unmodeled-tyler/vessel-browser
published an article about 2 hours ago
Article
Vessel Browser: The Open Source Browser Designed for Autonomous Agents
•
1
updated a
collection 8 days ago
replied to RakshitAralimatti's post 9 days ago
I've really been enjoying GLM-5.1 - It's what I've been using for the majority of my agent-based work these days. Absolutely zero complaints from me, and it got me off the $100/mo Claude Max plan so I call that a win lol
reacted to anakin87's post with ❤️ 9 days ago
Post
3285
📣 I just published a free course on Reinforcement Learning Environments for Language Models!
📌 COURSE: https://github.com/anakin87/llm-rl-environments-lil-course
Over the past year, we've seen a shift in LLM Post-Training.
Previously, Supervised Fine-Tuning was the most important part: making models imitate curated Question-Answer pairs.
Now we also have Reinforcement Learning with Verifiable Rewards. With techniques like GRPO, models can learn through trial and error in dynamic environments. They can climb to new heights without relying on expensively prepared data.
But what actually are these environments in practice❓ And how do you build them effectively❓
Fascinated by these concepts, I spent time exploring this space through experiments, post-training Small Language Models.
I've packaged everything I learned into this short course.
What you'll learn
🔹 Agents, Environments, and LLMs: how to map Reinforcement Learning concepts to the LLM domain
🔹 How to use Verifiers (open-source library by Prime Intellect) to build RL environments as software artifacts
🔹 Common patterns: How to build single-turn, multi-turn, and tool-use environments
🔹 Hands-on: turn a small language model (LFM2-2.6B by LiquidAI) into a Tic Tac Toe master
🔸 Build the game Environment
🔸 Use it to generate synthetic data for SFT warm-up
🔸 Group-based Reinforcement Learning
If you're interested in building "little worlds" where LLMs can learn, this course is for you.
---
🤗🕹️ Play against the trained model: anakin87/LFM2-2.6B-mr-tictactoe
📚 HF collection (datasets + models): https://huggingface.co/collections/anakin87/lfm2-26b-mr-tic-tac-toe
📌 COURSE: https://github.com/anakin87/llm-rl-environments-lil-course
Over the past year, we've seen a shift in LLM Post-Training.
Previously, Supervised Fine-Tuning was the most important part: making models imitate curated Question-Answer pairs.
Now we also have Reinforcement Learning with Verifiable Rewards. With techniques like GRPO, models can learn through trial and error in dynamic environments. They can climb to new heights without relying on expensively prepared data.
But what actually are these environments in practice❓ And how do you build them effectively❓
Fascinated by these concepts, I spent time exploring this space through experiments, post-training Small Language Models.
I've packaged everything I learned into this short course.
What you'll learn
🔹 Agents, Environments, and LLMs: how to map Reinforcement Learning concepts to the LLM domain
🔹 How to use Verifiers (open-source library by Prime Intellect) to build RL environments as software artifacts
🔹 Common patterns: How to build single-turn, multi-turn, and tool-use environments
🔹 Hands-on: turn a small language model (LFM2-2.6B by LiquidAI) into a Tic Tac Toe master
🔸 Build the game Environment
🔸 Use it to generate synthetic data for SFT warm-up
🔸 Group-based Reinforcement Learning
If you're interested in building "little worlds" where LLMs can learn, this course is for you.
---
🤗🕹️ Play against the trained model: anakin87/LFM2-2.6B-mr-tictactoe
📚 HF collection (datasets + models): https://huggingface.co/collections/anakin87/lfm2-26b-mr-tic-tac-toe
- 1 reply
reacted to chimbiwide's post with 🔥 9 days ago
Post
1529
We(@KeeganC and @chimbiwide ) have released our newest NPC roleplaying model: chimbiwide/Gemma4NPC-E4B , based on Gemma4-E4B with an improved training dataset.
We welcome any feedback and suggestions, please leave a comment!
We are releasing the E2B model soon, stay tuned!
We welcome any feedback and suggestions, please leave a comment!
We are releasing the E2B model soon, stay tuned!
- 1 reply
reacted to ArtelTaleb's post with ❤️ 12 days ago
Post
4783
🎵 MP3 Player - Drop your music, hit play. No install
MP3 Player - brings that energy back - straight in your browser.
- Drop your files - MP3, WAV, FLAC, AAC, OGG, AIFF, WMA — it reads them all
- Build your playlist - add tracks one by one or batch-load a whole folder
- Retro LCD display - scrolling track info, elapsed time, the full throwback
- Full controls - play, pause, skip, shuffle, repeat
- Mobile-first - big tactile buttons, works on phone like an iPod in your pocket
No install. No GPU needed on your end. Just upload and play.
👉 ArtelTaleb/mp3-player
MP3 Player - brings that energy back - straight in your browser.
- Drop your files - MP3, WAV, FLAC, AAC, OGG, AIFF, WMA — it reads them all
- Build your playlist - add tracks one by one or batch-load a whole folder
- Retro LCD display - scrolling track info, elapsed time, the full throwback
- Full controls - play, pause, skip, shuffle, repeat
- Mobile-first - big tactile buttons, works on phone like an iPod in your pocket
No install. No GPU needed on your end. Just upload and play.
👉 ArtelTaleb/mp3-player
reacted to allisonandreyev's post with 🔥 12 days ago
Post
5602
ConfCrawler 🕷️ — never miss a conference deadline again
Keeping track of submission deadlines across CV, NLP, robotics, and ML conferences is a mess. ConfCrawler aggregates them in one place so you can actually plan your research calendar.
What's in it:
- Deadlines for major conferences (CVPR, ICCV, NeurIPS, ICRA, ACL, etc.)
- Updated regularly
- Filterable by field / month
Built this out of personal frustration while juggling multiple submission cycles. Hope it saves someone else the tab-hoarding.
🔗 https://confcrawler.vercel.app/
feedback welcome — open to adding more conferences if yours isn't listed!
Keeping track of submission deadlines across CV, NLP, robotics, and ML conferences is a mess. ConfCrawler aggregates them in one place so you can actually plan your research calendar.
What's in it:
- Deadlines for major conferences (CVPR, ICCV, NeurIPS, ICRA, ACL, etc.)
- Updated regularly
- Filterable by field / month
Built this out of personal frustration while juggling multiple submission cycles. Hope it saves someone else the tab-hoarding.
🔗 https://confcrawler.vercel.app/
feedback welcome — open to adding more conferences if yours isn't listed!
reacted to alibidaran's post with 🔥 14 days ago
Post
4057
With the release of Gemma 4, I launched a new Space called MEDPAI — a medical imaging analysis tool that combines object detection with multimodal AI.
Here's how it works:
Upload a CT scan or X-ray
Computer vision models detect and annotate findings
Gemma 4 33B generates a report or answers your questions about the image
Currently available detectors: dental analysis and bone fracture detection.
More models are in the pipeline — follow the Space to stay updated!
alibidaran/MEDPAI
Here's how it works:
Upload a CT scan or X-ray
Computer vision models detect and annotate findings
Gemma 4 33B generates a report or answers your questions about the image
Currently available detectors: dental analysis and bone fracture detection.
More models are in the pipeline — follow the Space to stay updated!
alibidaran/MEDPAI
- 3 replies
replied to alibidaran's post 14 days ago
This is so cool! Awesome resource for people that might be interested in a bit more info about their results!
published a
model 16 days ago
liked 2
models 17 days ago
reacted to danielhanchen's post with 👀 17 days ago
Post
2695
A new way to use Unsloth.
Coming soon...
Coming soon...
reacted to qgallouedec's post with 🔥 17 days ago
Post
2300
TRL v1.0 is out!
Hugging Face's TRL library is downloaded 3 million times a month. Over 130k models trained with it are public on the Hub, and major projects like @unsloth and @axolotl-ai-co build directly on top of it. v1.0 is the moment we acknowledged that responsibility explicitly, with a real stability contract.
The field hasn't settled. Building stable software in a domain that keeps invalidating its own assumptions is the actual problem we're solving. The answer is a design that can absorb the next shift without breaking what people rely on.
What's in v1.0:
Deep Hugging Face integration, low infrastructure burden
What's next: asynchronous GRPO, better scaling support, and making training legible enough that agents can inspect and steer it.
Read more: hf.co/blog/trl-v1
Hugging Face's TRL library is downloaded 3 million times a month. Over 130k models trained with it are public on the Hub, and major projects like @unsloth and @axolotl-ai-co build directly on top of it. v1.0 is the moment we acknowledged that responsibility explicitly, with a real stability contract.
The field hasn't settled. Building stable software in a domain that keeps invalidating its own assumptions is the actual problem we're solving. The answer is a design that can absorb the next shift without breaking what people rely on.
What's in v1.0:
Deep Hugging Face integration, low infrastructure burden
What's next: asynchronous GRPO, better scaling support, and making training legible enough that agents can inspect and steer it.
pip install --upgrade trlRead more: hf.co/blog/trl-v1
reacted to sergiopaniego's post with 🔥 17 days ago
Post
1992
TRL is officially an adult 🥳
excited to announce TRL v1.0❗️
head to the blog to see how we got here and what’s next for this post-training library, designed to keep pace with the field
https://huggingface.co/blog/trl-v1
excited to announce TRL v1.0❗️
head to the blog to see how we got here and what’s next for this post-training library, designed to keep pace with the field
https://huggingface.co/blog/trl-v1
- 2 replies
reacted to alibidaran's post with 🔥 17 days ago
Post
2053
🧠 Introducing Qwen2.5 — Cognitive Reasoning Mode
I fine-tuned Qwen2.5 with GRPO to actually think before it answers — not just pattern-match.
Most LLMs mimic reasoning. This one builds a real cognitive path:
📌 Plan → understand the task
🔍 Monitor → reason step by step
✅ Evaluate → verify before answering
Every response follows a strict structured protocol:
<think> <planning> ... <monitoring> ... <evaluation> ... </think>
Then a clean, reasoning-free <output>.
The model self-checks its own structure. If a section is missing or malformed → the response is invalid.
This isn't chain-of-thought slapped on top. The reasoning protocol is baked in via RL.
🔗 Full README + inference code below 👇
alibidaran/Qwen_COG_Thinker_Merged
#AI #LLM #Qwen #ReasoningModels #GRPO #OpenSource
I fine-tuned Qwen2.5 with GRPO to actually think before it answers — not just pattern-match.
Most LLMs mimic reasoning. This one builds a real cognitive path:
📌 Plan → understand the task
🔍 Monitor → reason step by step
✅ Evaluate → verify before answering
Every response follows a strict structured protocol:
<think> <planning> ... <monitoring> ... <evaluation> ... </think>
Then a clean, reasoning-free <output>.
The model self-checks its own structure. If a section is missing or malformed → the response is invalid.
This isn't chain-of-thought slapped on top. The reasoning protocol is baked in via RL.
🔗 Full README + inference code below 👇
alibidaran/Qwen_COG_Thinker_Merged
#AI #LLM #Qwen #ReasoningModels #GRPO #OpenSource
- 2 replies
reacted to their post with 🚀 17 days ago
Post
2287
RESULTS ARE IN!
- Videos of each evaluation: https://www.youtube.com/playlist?list=PLkDBfeR-zsShiZ2HpcscFDH-36uDwsl5W
- Link to repo: https://github.com/unmodeled-tyler/vessel-browser
- quanta-intellect
Finally just wrapped up a comparative analysis of my new open source AI browser, Vessel, against Claude Chrome from Anthropic.
The test evaluates both web navigation harnesses for speed and efficiency on a simple real-world e-commerce task. Opus 4.6 was used for each of the 3 evaluations, and the results show that Opus 4.6 was AT LEAST 2X FASTER when using Vessel Browser for web navigation in place of Claude Chrome.
Results (in order, fastest to slowest)
1. Claude Code + Vessel Browser: 3 minutes and 10s
2. Hermes Agent + Vessel Browser: 4 minutes and 13s
3. Claude Code + Claude Chrome: 7 minutes and 57s
Vessel Browser is open source, designed explicitly for agents from the ground-up (it is not a fork of a human browser with AI features bolted on), and supports a local MCP server for agent control, or BYOK custom OAI endpoints. Check it out for yourself!
- Videos of each evaluation: https://www.youtube.com/playlist?list=PLkDBfeR-zsShiZ2HpcscFDH-36uDwsl5W
- Link to repo: https://github.com/unmodeled-tyler/vessel-browser
-
Finally just wrapped up a comparative analysis of my new open source AI browser, Vessel, against Claude Chrome from Anthropic.
The test evaluates both web navigation harnesses for speed and efficiency on a simple real-world e-commerce task. Opus 4.6 was used for each of the 3 evaluations, and the results show that Opus 4.6 was AT LEAST 2X FASTER when using Vessel Browser for web navigation in place of Claude Chrome.
Results (in order, fastest to slowest)
1. Claude Code + Vessel Browser: 3 minutes and 10s
2. Hermes Agent + Vessel Browser: 4 minutes and 13s
3. Claude Code + Claude Chrome: 7 minutes and 57s
Vessel Browser is open source, designed explicitly for agents from the ground-up (it is not a fork of a human browser with AI features bolted on), and supports a local MCP server for agent control, or BYOK custom OAI endpoints. Check it out for yourself!