AI & ML interests

None defined yet.

Recent Activity

mrfakenameย 
posted an update 2 days ago
view post
Post
408
Excited to share that I've joined the Hugging Face Fellows program! ๐Ÿค—

Looking forward to contributing to & working more closely with the open-source ecosystem - huge thanks to everyone who's supported me on this journey! ๐Ÿš€
Nymboย 
posted an update 11 days ago
view post
Post
4631
๐Ÿš€ I've just shipped a major update to the Nymbo/Tools MCP server: the Agent_Terminal, a single "master tool" that cuts token usage by over 90%!

Anthropic found 98.7% context savings using code execution with MCP, Cloudflare published similar findings. This is my open-source implementation of the same idea.

# The Problem

Traditional MCP exposes every tool definition directly to the model. With 12 tools, that's thousands of tokens consumed *before the conversation even starts*. Each tool call also passes intermediate results through the context window โ€” a 10,000-row spreadsheet? That's all going into context just to sum a column.

# The Solution: One Tool to Rule Them All

Agent_Terminal wraps all 12 tools (Web_Search, Web_Fetch, File_System, Generate_Image, Generate_Speech, Generate_Video, Deep_Research, Memory_Manager, Obsidian_Vault, Shell_Command, Code_Interpreter) into a single Python code execution gateway.

Instead of the model making individual tool calls, it writes Python code that orchestrates the tools directly:

# Search for Bitcoin price
result = Web_Search("current price of bitcoin", max_results=3)
print(result)


Don't know what tools are available? The agent can discover them at runtime:

print(search_tools('image'))  # Find tools by keyword
print(usage('Generate_Image'))  # Get full docs for a specific tool


The individual direct tool calls are all still there, but they can be disabled if using the Agent_Terminal. Try it now - https://www.nymbo.net/nymbot
  • 1 reply
ยท
JingzeShiย 
posted an update 26 days ago
Nymboย 
posted an update about 1 month ago
view post
Post
893
I've added an 11th tool to the Nymbo/Tools MCP server, it's for your Obsidian_Vault. I'd argue it's far more context-efficient than any other Obsidian MCP I've seen, and doesn't require any plugins. Also some big improvements to the Web_Search and Web_Fetch tools.

# Obsidian_Vault Tool

It's basically a read-only version of the File_System tool, but it works so well for navigating Obsidian without unnecessary context. It supports recursive (full-text) search across the entire vault, and supports offset so the agent can "scroll" through a document without re-consuming tokens.

Run the server locally and set the OBSIDIAN_VAULT_ROOT environment variable to your vault's root path. If you don't use Obsidian, this is perfectly usable as simply a read-only filesystem.

# Web_Search Improvements

The Web_Search tool previously just used DuckDuckGo as a backend search engine, but now it also supports Bing, Brave, Yahoo, and Wikipedia. Default engine is auto which provides results from all backends in recommended order. Still doesn't require any kind of API or auth for Web_Search.

There's also a new date filter to limit results to those created in the past day, week, month, or year. Oh, and uhh, SafeSearch is now off by default :)

# Web_Fetch Improvements

As context-efficient as the Markdown mode is for web browsing, sometimes it does lose important context in the conversion from HTML to Markdown. So I've added a new HTML mode to the Web_Fetch tool that basically executes a cURL request on the URL, returning the full HTML page if necessary.

# A Note on Claude Skills

I've been having fun with the new File_System and Shell_Command tools. Using Claude Skills doesn't currently work in the public HF space because of environment restrictions, but using Skills works perfectly well running locally.

Happy building ~
megย 
posted an update about 1 month ago
view post
Post
3727
๐Ÿค– Did you know your voice might be cloned without your consent from just *one sentence* of audio?
That's not great. So with @frimelle , we brainstormed a new idea for developers who want to curb malicious use: โœจThe Voice Consent Gate.โœจ
Details, code, here: https://huggingface.co/blog/voice-consent-gate
  • 3 replies
ยท
mrfakenameย 
posted an update about 1 month ago
view post
Post
5970
Trained a model for emotion-controllable TTS based on MiMo audio on LAION's dataset.

Still very early and does have an issue with hallucinating but results seem pretty good so far, given that it is very early into the training run.

Will probably kick off a new run later with some settings tweaked.

Put up a demo here: https://huggingface.co/spaces/mrfakename/EmoAct-MiMo

(Turn ๐Ÿ”Š on to hear audio samples)
ยท
Nymboย 
posted an update about 2 months ago
view post
Post
1965
Two new tools added to the Nymbo/Tools MCP server, File_System and Shell_Exec. You can theoretically do basically anything with these two tools, and it should enable support for many Claude Skills.

GPT-5-Codex proves that for many cases, shell commands really are all you need, and Claude Skills seem to lean into this. The thing is, nothing about the design of Claude Skills actually restricts them to proprietary models!

# File_System

There's a new directory inside the repo called Filesystem, that's the agent's "root". It can perform the following actions : list, read, write, append, mkdir, move, copy, delete, info, help. It's able to keep this all within the scope of one tool call by making the Action field required and all other fields optional. Using a filesystem shouldn't require 15 different tools.

Files created in the public HF space live in the space's running container, and gets cleared when the space is restarted. When running the server locally, files are actually stored on disk.

# Shell_Exec

What good is a filesystem if you can't execute commands in that filesystem? This tool automatically detects if the server is running on Windows or Linux, and suggests using the appropriate shell (PowerShell/Bash). Both of these new tools require that the agent uses relative paths, rather than absolute paths. I could be convinced to back pedal on this.

# Closing Thoughts

The File_System and Shell_Exec tools aren't super polished yet, I'll continue to improve the agent's instructions and UX of using the new tools. Most of my testing was done with gpt-oss-20b and if it messes up, it gets the gist after one failed tool call. It should work perfectly fine for the GPU poor.
  • 1 reply
ยท
m-ricย 
posted an update about 2 months ago
view post
Post
747
Tokenization is one of the most important processes in AI - yet many would like to kill it ๐Ÿ’€

What's tokenization? The neural networks inside LLMs actually only process numbers, not text: tokenization is the process that makes text readable for them, by converting sentences into lists of numbers.

โžก๏ธ For instance, "This is tokenization" would be split into "This | is | token | ization", then each of the parts (tokens) are converted to IDs according to a predefined mapping: for instance "ization" could map to id 2438.
Thus "This is tokenization" can become 1335 | 135 | 2980 | 2438 => now the model can process the sentence!

Most tokenizers today use pre-specified mappings called "vocabularies", generally built about the compression algorithme Byte-Pair Encoding (BPE) that learns from a big corpuses of texts an optimized split to efficiently encode any text from the same distribution into a list token IDs.

๐Ÿคจ Now, these current tokenizers have flaws.
For instance, the rigidity of their mapping creates losses ; the prime example being that a tokenizer designed for English (thus optimized for tokens like "has", "been", "clock", etc) will not have the right tokens to approach Burmese, thus being terribly inefficient at it.

Many alternative approaches have emerged as a result: for instance "tokenizer-free tokenizers". One that I really liked was "entropy-based": it monitors the stream of text, and trigger a split whenever the entropy increases too much, i.e. when something "surprising" happens.

But this great article argues that tokenizers are a lesser evil. Read and decide for yourself!
https://huggingface.co/blog/catherinearnett/in-defense-of-tokenizers
Nymboย 
posted an update about 2 months ago
view post
Post
1836
I've made some improvements to my custom Deep_Research tool in the Nymbo/Tools MCP server. I've added a second LLM process and it still takes less than 1 minute to complete!

The original version of my Deep_Research tool would basically dump up to 50 fetched webpages onto the Researcher model (Qwen3-235B), with only a little bit of context shown from each page.

# New "Filterer" Process

The new process includes another LLM call before the researcher process. The Filterer (also Qwen3-235B) gets the query summary and the original 50 pages with low context, and decides which pages are most relevant to the research topic. The Filterer then outputs the URLs to the relevant pages, which are then re-fetched (with more context) and sent to the Researcher.

# Researcher Context

The Researcher now gets only the relevant webpages, then begins writing the report. When testing with 50 initial results, the researcher would often end up with 10-20 results of relevant context.

It still takes less than a minute to accomplish everything, thanks entirely to Cerebras inference. It now takes about 35-45 seconds to complete once the tool is run.

It's also worth noting that both the Filterer and Researcher now are provided the current time/date before they see the content, reducing hallucinations caused by knowledge cutoffs.
hassenhamdiย 
posted an update about 2 months ago
view post
Post
205
New release of HyDRA v0.2 is here!

๐Ÿ HyDRA: Hybrid Dynamic RAG Agent.

For addressing the limitations of simple, static RAG. HyDRA is the answer. It's an advanced, unified framework for agentic RAG, inspired by the latest research to create something truly powerful.

๐Ÿง  Moving beyond single-shot retrieval. HyDRA introduces a multi-turn, reflection-based system with coordinated agents: a Planner, Coordinator, and Executors (currently local & deep web search).

๐Ÿ”ฌ At its core is an advanced 3-stage local retrieval pipeline that leaves basic RAG in the dust:
๐Ÿฅ‡ 1. Hybrid Search: Combines dense (semantic) and sparse (textual) embeddings in one go using the bge-m3 model. This alone is a massive upgrade.
๐Ÿฅˆ 2. RRF (Reciprocal Rank Fusion): Intelligently merges and reranks results from different search vectors for ultimate precision.
๐Ÿฅ‰ 3. Advanced Reranking: Uses the bge-m3-reranker model to score and surface the absolute most relevant documents for any query.

โšก๏ธ This isn't just powerful, it's blazing fast. We're using SOTA ANN (HNSW) with vector and index quantization (down to 1-bit!) for near-instant retrieval with minimal quality loss.

๐Ÿค– HyDRA is more than just retrieval. It incorporates memory from experience and reflection, creating a guiding policy for smarter future interactions and strategic planning.

The result? A local retrieval system that significantly outperforms standard vector search RAG.

๐ŸŒ For deep web searches, HyDRA leverages the asynDDGS library and mcp (Model Context Protocol) for free, unrestricted web access. The entire reasoning engine is powered by the incredibly fast and efficient Google Gemini 2.5 Flash!

๐Ÿ‘จโ€๐Ÿ’ป Explore the project, dive into the code, and see it in action:
๐Ÿ”— GitHub: https://github.com/hassenhamdi/HyDRA (leave a star if you like the project)

๐Ÿค Looking to implement cutting-edge AI solutions or collaborate? Let's connect!
LinkedIn: linkedin.com/in/hassenhamdi
Email: [email protected]
Discord: hassenhamdi
m-ricย 
posted an update about 2 months ago
view post
Post
4877
STOP EVERYTHING NOW - we might finally have a radical architecture improvement over Transformers!!! ๐Ÿšจ

A lone scientist just proposed Tiny Recursive Model (TRM), and it is literally the most impressive model that I've seen this year.

โžก๏ธ Tiny Recursive Model is 7M parameters
โžก๏ธ On ARC-AGI, it beats flagship models like Gemini-2.5-pro

Consider how wild this is: Gemini-2.5-pro must be over 10,000x bigger
and had 1,000 as many authors ๐Ÿ˜‚ (Alexia is alone on the paper)

What's this sorcery?
In short: it's a very tiny Transformers, but it loops over itself at two different frequencies, updating two latent variables: one for the proposed answer and one for the reasoning.

@AlexiaJM started from the paper Hierarchical Reasoning Model, published a few months ago, that already showed breakthrough improvement on AGI for its small size (27M)

Hierarchical Reasoning Model had introduced one main feature:
๐Ÿ”Ž Deep supervision
In their model, one part (here one layer) would run at high frequency, and another would be lower frequency, running only every n steps.

They had used a recurrent architecture, where these layers would repeat many times ; but to make it work they had to do many approximations, including not fully backpropagating the loss through all layers.

Alexia studied what was useful and what wasn't, and cleaned the architecture as follows :
Why use a recurrent architecture, when you can just make it a loop?
โžก๏ธ She made the network recursive, looping over itself

Why use 2 latent variables ?
โžก๏ธ She provides a crystal clear explanation : the one that changes frequently is the reasoning, the one that changes at low frequency is the proposed answer.
โžก๏ธ She runs ablation studies to validate that 2 is indeed optimal.

This new setup is a much more elegant way to process reasoning than generating huge chains of tokens as all flagship models currently do.

This might be the breakthrough we've been awaiting for so long!
ยท
Nymboย 
posted an update 2 months ago
view post
Post
674
I have a few Sora-2 invites - 15509N
  • 1 reply
ยท
Tonicย 
posted an update 3 months ago
Nymboย 
posted an update 3 months ago
view post
Post
1101
There's now a custom Deep_Research tool in my Nymbo/Tools MCP server! TL;DR: The agent using the tools writes a summary of your requests and up to five DuckDuckGo searches (up to 50 results). Each of the webpages found in the searches are then fetched and given to our researcher (Qwen3-235B-A22B-Thinking-2507). The researcher sees the summary, searched queries, and fetched links, then writes a thorough research report. The agent using the tool provides the user with a summary of the report and a link to download research_report.txt. The researcher's instructions are similar to some leaked Perplexity sys prompts.

# Deep_Research Tool

It accomplishes everything in under a minute so it doesn't hit MCP's 60 second timeout, mostly thanks to Cerebras. The only thing required to make this work is a HF_READ_TOKEN for inference.

The Deep_Research tool could certainly be improved. It still needs some sort of mechanism for sorting URLs based on importance (I've got some ideas but I don't want it to be the responsibility of the agent using the tool). I'll probably add a second researcher to filter out the bad sources before inferencing the big researcher. I'm hellbent on keeping this all within the scope of one tool call.

# More Fetch/Web Search Improvements

The Search_DuckDuckGo tool has been further enhanced. It now allows the agent to browse through all pages of results. The results also now include published date (if detected). It also now supports every DDG search types! Default DDG search is called text, but it can also now search by news, images, videos, and books.

The Fetch_Webpage tool now specifies how much of the page has been truncated, and cursor index, allowing it to pickup where it left off without re-consuming tokens. The model can now also choose to strip CSS selectors to remove excess noise, and there's a new URL Scraper mode that only returns URLs found on the full page.

More to come soon ~
megย 
posted an update 3 months ago
view post
Post
2892
๐Ÿค– As AI-generated content is shared in movies/TV/across the web, there's one simple low-hanging fruit ๐Ÿ‡ to help know what's real: Visible watermarks. With the Gradio team, I've made sure it's trivially easy to add this disclosure to images, video, chatbot text. See how: https://huggingface.co/blog/watermarking-with-gradio
Thanks to the code collab in particular from @abidlabs and Yuvraj Sharma.
Tonicย 
posted an update 3 months ago
view post
Post
762
COMPUTER CONTROL IS ON-DEVICE !

๐Ÿก๐Ÿค– 78 % of EU smart-home owners DONโ€™T trust cloud voice assistants.

So we killed the cloud.

Meet Extรฉ: a palm-sized Android device that sees, hears & speaks your language - 100 % offline, 0 % data sent anywhere.

๐Ÿ”“ We submitted our technologies for consideration to the Liquid AI hackathon.

๐Ÿ“Š Dataset: 79 k UI-action pairs on Hugging Face (largest Android-control corpus ever) Tonic/android-operator-episodes

โšก Model: 98 % task accuracy, 678MB compressed , fits on existing android devices ! Tonic/l-android-control

๐Ÿ›ค๏ธ Experiment Tracker : check out the training on our TrackioApp Tonic/l-android-control

๐ŸŽฎ Live Model Demo: Upload an Android Screenshot and instructions to see the model in action ! Tonic/l-operator-demo



Built in a garage, funded by pre-orders, no VC. Now weโ€™re scaling to 1 k installer units.

Weโ€™re giving 50 limited-edition prototypes to investors , installers & researchers who want to co-design the sovereign smart home.

๐Ÿ‘‡ Drop โ€œEUSKERAโ€ in the comments if you want an invite, tag a friend who still thinks Alexa is โ€œconvenient,โ€ and smash โ™ฅ๏ธ if AI should belong to people - not servers.
ยท
Nymboย 
posted an update 3 months ago
view post
Post
1051
I have a few updates to my MCP server I wanna share: New Memory tool, improvements to web search & speech generation.

# Memory_Manager Tool

We now have a Memory_Manager tool. Ask ChatGPT to write all its memories verbatim, then tell gpt-oss-20b to save each one using the tool, then take them anywhere! It stores memories in a memories.json file in the repo, no external database required.

The Memory_Manager tool is currently hidden from the HF space because it's intended for local use. It's enabled by providing a HF_READ_TOKEN in the env secrets, although it doesn't actually use the key for anything. There's probably a cleaner way of ensuring memory is only used locally, I'll come back to this.

# Fetch & Websearch

The Fetch_Webpage tool has been simplified a lot. It now converts the page to Markdown and returns the page with three length settings (Brief, Standard, Full). This is a lot more reliable than the old custom extraction method.

The Search_DuckDuckGo tool has a few small improvements. The input is easier for small models to get right, and the output is more readable.

# Speech Generation

I've added the remaining voices for Kokoro-82M, it now supports all 54 voices with all accents/languages.

I also removed the 30 second cap by making sure it computes all chunks in sequence, not just the first. I've tested it on outputs that are ~10 minutes long. Do note that when used as an MCP server, the tool will timeout after 1 minute, nothing I can do about that for right now.

# Other Thoughts

Lots of MCP use cases involve manipulating media (image editing, ASR, etc.). I've avoided adding tools like this so far for two reasons:

1. Most of these solutions would require assigning it a ZeroGPU slot.
2. The current process of uploading files like images to a Gradio space is still a bit rough. It's doable but requires additional tools.

Both of these points make it a bit painful for local usage. I'm open to suggestions for other tools that rely on text.
Tonicย 
posted an update 3 months ago
view post
Post
712
๐Ÿ™‹๐Ÿปโ€โ™‚๏ธ Hey there folks ,

Just wanted to annouce ๐ŸญSmolFactory : it's the quickest and best way to finetune SmolLM3 and GPT-OSS-20B on huggingface !

Basicaly it's an app you can run on huggingface by duplicating the space and running your training directly on huggingface GPUs .

It will help you basically select datasets and models, fine tune your model , make an experiment tracker you can use on your mobile phone , push all your model card and even automatically make a demo for you on huggingface so you can directly test it out when it's done !

check out the blog to learn more : https://huggingface.co/blog/Tonic/smolfactory

or just try the app directly :
Tonic/SmolFactory

you can vibe check the cool models I made :
French SmolLM3 : Tonic/Petite-LLM-3
Medical GPT-OSS : Tonic/med-gpt-oss-20b-demo

check out the model cards :
multilingual reasoner (gpt-oss) - Tonic/gpt-oss-20b-multilingual-reasoner
med-gpt-oss : Tonic/med-gpt-oss-20b
petite-elle-l-aime : Tonic/petite-elle-L-aime-3-sft

github repo if you like command line more than gradio : https://github.com/josephrp/smolfactory

drop some likes on these links it's really much appreciated !

feedback and PRs are welcome !
Nymboย 
posted an update 3 months ago
view post
Post
1029
I built a general use MCP space ~ Fetch webpages, DuckDuckGo search, Python code execution, Kokoro TTS, Image Gen, Video Gen.

# Tools

1. Fetch webpage
2. Web search via DuckDuckGo (very concise, low excess context)
3. Python code executor
4. Kokoro-82M speech generation
5. Image Generation (use any model from HF Inference Providers)
6. Video Generation (use any model from HF Inference Providers)

The first four tools can be used without any API keys whatsoever. DDG search is free and the code execution and speech gen is done on CPU. Having a HF_READ_TOKEN in the env variables will show all tools. If there isn't a key present, The Image/Video Gen tools are hidden.

Nymbo/Tools
  • 1 reply
ยท
Nymboย 
posted an update 4 months ago
view post
Post
1047
Anyone using Jan-v1-4B for local MCP-based web search, I highly recommend you try out Intelligent-Internet/II-Search-4B

Very impressed with this lil guy and it deserves more downloads. It's based on the original version of Qwen3-4B but find that it questions reality way less often. Jan-v1 seems to think that everything it sees is synthetic data and constantly gaslights me