Hugging Face's logo

Join the conversation

Join the community of Machine Learners and AI enthusiasts.

Sign Up

All HF Hub posts
DavidAU 
posted an update 2 days ago
view post
Post
6305
Qwen3.8 - 27B - COLD FUSION

Reduction in thinking tokens to 1/10 to 1/2 "normal Qwen" across all three modes of thinking without loss of detail in thinking or output.

Increase in general intelligence too (all 7 critical benches); bench marks posted.

Trained using the COLD FUSION method (GAIN + UNSLOTH).
NOTE: This is NOT a heretic / uncensored version.

MTP and GGUF NEO MAX quants:
DavidAU/Qwen3.8-27B-Cold-Fusion-GAIN-V1.1-NM-DAU-NEO-MAX-MTP-GGUF

SOURCE:
DavidAU/Qwen3.8-27B-Cold-Fusion-GAIN-V1.1
  • 3 replies
·
SeaWolf-AI 
posted an update 2 days ago
view post
Post
3531
3,631 candidate molecules arrived in five days, from 83 accounts — roughly 700 a day. Far more than we expected. Thank you.

Yesterday we opened the third season and 224 arrived within a day: Chagas disease.

Why this disease

Around 6 million people live with it, mostly in Latin America (WHO). Many carry it for decades without knowing, while the heart is slowly damaged. There are two drugs and both date from the 1960s, hard enough to tolerate that many patients cannot finish the two-month course.

Sixty years without a new drug is not only a scientific problem. Most patients live where development costs cannot be recovered, which is why WHO calls this a neglected tropical disease.

But the cost of proposing a candidate and filtering it has changed. So it seemed worth asking whether work nobody funds could be done by many people sharing it out.

The problem this season

The target is CYP51, the enzyme T. cruzi uses to build its membrane sterols. Block it and the parasite cannot survive. The difficulty is that we carry the same enzyme.

Scoring: binding 30 · selectivity 30 · ADMET 15 · whole-cell 10 · novelty 10 · synthesis 5

Selectivity carries 30 points because nobody has solved it. Among the approved azoles on the board as reference compounds, some score 0 on selectivity — not a scorer fault, but the measurement.

Taking part

Design with any model, submit a SMILES, scored within minutes. Five ready-to-paste prompts per season, and the full rubric is published. Your molecule stays yours; private submission is the default.

Prizes — 4,000 USD across three seasons

Malaria 30 Sep · 1,000 | Tuberculosis 31 Oct · 2,000 | Chagas 30 Nov · 1,000

We know this does not cover the time you spend. It is a way of saying the work had worth.

FINAL-Bench/open-discovery-challenge
Enderchef 
posted an update about 19 hours ago
ProCreations 
posted an update 3 days ago
view post
Post
3293
I have hit 300 followers, and I think this calls for a bit of a giveaway 👀 a unique one, too.
I have had countless AI projects I have wanted to make but have been (brutally) blocked by compute. Now that I finally have just enough compute to sort of get around (i still don't have enough 😭) and for hitting 300 followers (tysm!) I will be funding three of the communities projects via HuggingFace jobs, giving them 150 dollars max worth of compute each. I will personally be picking the winners, I am looking for projects that genuinely hit the compute wall: great ideas, blocked by compute, just like the countless ideas I've had.
To join, head over to https://giveaway.ssh.codes
RULES:
- Final result must be open weight or open source
- Only one submission per person
- Have fun!
  • 20 replies
·
Felladrin 
posted an update 2 days ago
view post
Post
4013
I've open-sourced the trainer I've been using to build tiny language models from scratch, together with the 95M base model I trained with it.

The trainer runs on Deno (https://deno.com, cross-platform), trains on WebGPU, and it writes GGUF directly. No Python/PyTorch. The weights live in a GGUF file from the first step to the last, so every checkpoint is already something llama.cpp can load.

The model is Felladrin/Minueza-3-95M-Base: 94.7M parameters, 1.95B tokens seen, 8192 context.

And here’s the repository on GitHub: https://github.com/felladrin/gguf-trainer

Here on Hugging Face, I published the optimizer state next to the weights, so you can continue the pretraining instead of starting over.

Or start your own from nothing: deno run -A cli.ts demo trains a tiny one end to end in under a minute.

And the docs are written for coding agents, so you can point your agent of choice at the GitHub repo and have it drive the whole pipeline.
  • 36 replies
·
Banaxi-Tech 
posted an update 1 day ago
view post
Post
2107
We're releasing BananaMind 2.1 Unified, a 35M three-tower relay model where the two output towers can only talk to each other through a silent middle tower that has no output head and no loss term.

Tower B trains entirely on indirect gradient. It was never told what to predict. It woke up anyway. Jacobian lens shows it carrying the correct answer ("Paris", "oxygen", "blue") at its deepest layer. Its bridge gates grew 4-47x from init. Feed it from only one side and the representations collapse to junk — it needs both outer towers to become semantic.

The PIQA result is the cleanest demonstration: Tower A alone scores 50.11 (chance). Tower C alone 52.12. Full system 61.75. All physical reasoning lives in the integration. The 35M three-tower beats the 50M single-tower BananaMind 2 Medium on PIQA.

Trained on 38B tokens in ~7.5 hours on 8x RTX PRO 6000. Ships with 7 ablation modes so you can surgically cut the model apart without retraining. Full training logs, J-lens fits, and eval outputs for every mode included.

This is a research model. It is interesting because you can take it apart.
BananaMind/BananaMind-2.1-Unified

Follow us for more:
BananaMind

@Banaxi-Tech
  • 4 replies
·
prithivMLmods 
posted an update 2 days ago
view post
Post
4296
The Qwen3.8 27B demo for object grounding is now available on Hugging Face Spaces.

It features three tasks: Object Detection (Bounding Boxes), Point Localization (Keypoints), and Spatial Guidance (Path Mapping).

Try it now: prithivMLmods/Qwen3.8-27B-Object-Detection
nightmedia 
posted an update 1 day ago
view post
Post
1887
Qwen3.8-27B metrics

It's hard to track all model cards where I post these, so I figured people would get more value out of seeing these in the open. 

quant     arc   arc/e boolq hswag obkqa piqa  wino
mxfp8     0.591,0.782,0.896,0.746,0.448,0.801,0.711
q8-hi     0.602,0.779,0.896,0.747,0.446,0.793,0.703
q6-hi     0.602,0.775,0.895,0.748,0.448,0.795,0.710
q4-hi     0.604,0.780,0.898,0.744,0.454,0.795,0.708
mxfp4     0.581,0.771,0.889,0.738,0.442,0.798,0.713
1M
mxfp8     0.590,0.787,0.897,0.744,0.446,0.801,0.709

Quant     Perplexity      Peak Memory   Tokens/sec
mxfp8     6.090 ± 0.054   34.74 GB      138
mxfp4     5.952 ± 0.051   21.30 GB      148

The performance is as measured on a M4 MBP 128GB, speed may vary depending on your platform.

These are all instruct metrics, generated by including this line in the jinja template:

{%- set enable_thinking = false %}


Then run the test suite to generate the metrics:
mlx_lm.evaluate --model MODEL --tasks winogrande boolq arc_challenge arc_easy hellaswag openbookqa piqa

This will generate the file:
eval_MODEL_0.4.9_winogrande_boolq_arc_challenge_arc_easy_hellaswag_openbookqa_piqa

This is a JSON containing all gathered metrics; for example the q4-hi:
"arc_challenge": {
        "alias": "arc_challenge",
        "acc,none": 0.5819112627986348,
        "acc_stderr,none": 0.014413988396996116,
        "acc_norm,none": 0.6040955631399317,
        "acc_norm_stderr,none": 0.01429122839353657
    },

I use the value of acc_norm for metrics, rounded to 3 decimals.

As I get more quants tested, I will add them here.

A complete test run for a single quant takes 7-9 hours depending on quant size, 10-12 hours for BF16 depending on perplexity: this is why you see on my model cards that I usually post the first three, that only take 2-3 hours :)

-G
  • 7 replies
·
tomaarsen 
posted an update 2 days ago
view post
Post
3443
🚨 I've just published Sentence Transformers v6.0, introducing MultiVectorEncoder: ColBERT-style late interaction models are now a fourth model type, for training, inference, and interpretation, alongside the dense, sparse, and reranker models! Details:

Where a regular embedding model compresses a whole text into one vector, a multi-vector model keeps one vector per token and scores query against document with the MaxSim operator. That preserves token-level matching information that a single vector has to average away. It is also the state of the art for visual document retrieval, where a text query is matched against page images directly, charts and tables included, with no OCR step in between.

Any PyLate, Stanford ColBERT, or ColPali checkpoint loads straight into the same familiar API: model.encode_query(), model.encode_document(), and model.similarity() just work, whether the documents are texts or page images.

Does it help? LightOn trained LateOn (multi-vector) and DenseOn (dense) on the same data with the same 149M ModernBERT backbone, and the multi-vector model wins on 9 of the 13 NanoBEIR datasets: 0.6868 vs 0.6764 mean NDCG@10. The price is a bigger index, and the new HierarchicalTokenPooling module halves it at roughly no retrieval cost.

Antoine Chaffin, Raphaël Sourty, and I wrote a blog post walking through multi-vector models in practice: loading the various checkpoint formats, encoding and scoring, plugging them into a search stack, running them on page images, and keeping the index affordable. Check it out if you want to get started, or just point your Agent to the URL: https://huggingface.co/blog/multi-vector-encoder

pip install sentence-transformers==6.0.0

Release notes: https://github.com/huggingface/sentence-transformers/releases/tag/v6.0.0
Bc-AI 
posted an update about 13 hours ago
Recently active users