Post
4298
I am LOVING HuggingFace Pro <3
- 7 replies
Daniel B. PRO
FlameF0X
P(doom) 1%
AI & ML interests
Pre-training text generator.
(Brother, im 18)
Please don't try to contact me.
Recent Activity
updated a model about 4 hours ago
FlameF0X/smoll-d1 published a model about 4 hours ago
FlameF0X/smoll-d1 upvoted a changelog about 5 hours ago
Contributor permission for access tokensOrganizations
reacted to mithulaartigala's post with 🔥 about 6 hours ago
reacted to AtAndDev's post with 👀 1 day ago
Post
4897
@Banaxi-Tech stop hiding my comments. AND STOP STEALING PAPERS AND SPREADING MISINFORMATION.
your BGA blog is a copy of NSA (deepseek, 2025) branded under your name. literally the same top16 selected blocks, 512 local window, router over block summaries, all you did was change block size from 64 to 128.
you didnt cite NSA once but you put a “please cite BGA” bibtex at the bottom.
i commented under your post and said that there is no way that you can support claims like: “The Accuracy Should BE WAy better than DSA but untested yet.” you didnt run a single experiment. and the 256x isnt from BGA, its just n/2k with k=2048 so the exact same k DSA uses. if opus wrote this for you, at least read it before posting.
i commented again after you hid my comment despite it having constructive and correct feedback and you hid that too. and again.
you can hide the truth and just try to get hf post likes..... but is it really the thing that needs to be done? do you really want to take papers and make them yours while barely even changing the params?
admitting your mistakes and doing something about them needs humbleness, intelligence, humanness.
i encourage you to admit your mistakes and try to do better next time (at least read what blog your ai wrote or do proper experiments to back your stuff up).
your BGA blog is a copy of NSA (deepseek, 2025) branded under your name. literally the same top16 selected blocks, 512 local window, router over block summaries, all you did was change block size from 64 to 128.
you didnt cite NSA once but you put a “please cite BGA” bibtex at the bottom.
i commented under your post and said that there is no way that you can support claims like: “The Accuracy Should BE WAy better than DSA but untested yet.” you didnt run a single experiment. and the 256x isnt from BGA, its just n/2k with k=2048 so the exact same k DSA uses. if opus wrote this for you, at least read it before posting.
i commented again after you hid my comment despite it having constructive and correct feedback and you hid that too. and again.
you can hide the truth and just try to get hf post likes..... but is it really the thing that needs to be done? do you really want to take papers and make them yours while barely even changing the params?
admitting your mistakes and doing something about them needs humbleness, intelligence, humanness.
i encourage you to admit your mistakes and try to do better next time (at least read what blog your ai wrote or do proper experiments to back your stuff up).
- 49 replies
reacted to danielhanchen's post with ❤️ 3 days ago
Post
4519
You can now train your own Decision model like Jev locally!
We increased Qwen3.5 0.8B’s aggregate accuracy from 20.7% to 74.3% across 3 decision benchmarks - on just 4GB VRAM.
Turn any LLM like Qwen3.8, Gemma 4 into decision models with our open-source Unsloth repo.
GitHub: https://github.com/unslothai/unsloth
Guide: https://unsloth.ai/docs/basics/train-your-own-decision-model-with-unsloth
We increased Qwen3.5 0.8B’s aggregate accuracy from 20.7% to 74.3% across 3 decision benchmarks - on just 4GB VRAM.
Turn any LLM like Qwen3.8, Gemma 4 into decision models with our open-source Unsloth repo.
GitHub: https://github.com/unslothai/unsloth
Guide: https://unsloth.ai/docs/basics/train-your-own-decision-model-with-unsloth
- 5 replies
replied to their post 20 days ago
I dont see a issue if them if its clear that they're bots
you... you don't?
reacted to TobiasLogic's post with 🔥 21 days ago
Post
3570
We’ve been cooking something new at Bench Labs.
Introducing Cagliostro-v3, our new 146M parameter language model trained completely from scratch.
The run isn’t even finished yet.
At the current checkpoint:
• 146M parameters
• 72.7B / 75B tokens trained
• 26.27 Open SLM Index
• 43.80 ArithMark-3
• Trained on a single RTX 5090
• ~90K to 103K tokens/sec during training
• ~9 days for the full run
• Apache 2.0
For some context, SmolLM2-135M scores 27.13 on the same Index after being trained on roughly 2 trillion tokens.
Cagliostro-v3 is currently at 26.27 with only ~72.7B.
That’s around 27x fewer training tokens.
The model also currently Hold the number 3rd spot for ArithMark-3, scoring 43.80
This wasn’t achieved by just throwing more tokens at the model. A huge part of v3 has been figuring out architecture, data mixture, and training dynamics at this scale.
The model uses a custom 30-layer decoder architecture with grouped-query attention and cross-head subspace attenuation, SwiGLU, RMSNorm, RoPE, tied embeddings, and a warmup-stable-decay training schedule.
During cooldown we also substantially shifted the data mixture toward higher-quality synthetic textbook and mathematics data, with the mathematics share increasing from 10% to 28%.
And everything is open.
The repository contains the training history with checkpoints pushed roughly every 30 minutes, so you can inspect how the model evolved throughout training rather than only seeing the final weights.
This is still a pre-final checkpoint. We have roughly 2.3B tokens left and the learning-rate cooldown is still running.
So 26.27 isn’t the final number.
Really excited to see where the last part of the run lands.
Cagliostro-v3:
bench-labs/cagliostro-v3
Built by Bench Labs.
Open SLM Leaderboard:
AxiomicLabs/Open_SLM_Leaderboard
Introducing Cagliostro-v3, our new 146M parameter language model trained completely from scratch.
The run isn’t even finished yet.
At the current checkpoint:
• 146M parameters
• 72.7B / 75B tokens trained
• 26.27 Open SLM Index
• 43.80 ArithMark-3
• Trained on a single RTX 5090
• ~90K to 103K tokens/sec during training
• ~9 days for the full run
• Apache 2.0
For some context, SmolLM2-135M scores 27.13 on the same Index after being trained on roughly 2 trillion tokens.
Cagliostro-v3 is currently at 26.27 with only ~72.7B.
That’s around 27x fewer training tokens.
The model also currently Hold the number 3rd spot for ArithMark-3, scoring 43.80
This wasn’t achieved by just throwing more tokens at the model. A huge part of v3 has been figuring out architecture, data mixture, and training dynamics at this scale.
The model uses a custom 30-layer decoder architecture with grouped-query attention and cross-head subspace attenuation, SwiGLU, RMSNorm, RoPE, tied embeddings, and a warmup-stable-decay training schedule.
During cooldown we also substantially shifted the data mixture toward higher-quality synthetic textbook and mathematics data, with the mathematics share increasing from 10% to 28%.
And everything is open.
The repository contains the training history with checkpoints pushed roughly every 30 minutes, so you can inspect how the model evolved throughout training rather than only seeing the final weights.
This is still a pre-final checkpoint. We have roughly 2.3B tokens left and the learning-rate cooldown is still running.
So 26.27 isn’t the final number.
Really excited to see where the last part of the run lands.
Cagliostro-v3:
bench-labs/cagliostro-v3
Built by Bench Labs.
Open SLM Leaderboard:
AxiomicLabs/Open_SLM_Leaderboard
- 8 replies
posted an update 21 days ago
Post
4647
Personally, I don't think bot accounts on Hugging Face are a good thing, as we don't know how many accounts are run by automated systems versus how many actual users there are. Dead Internet theory is already a thing.
To clarify I am taking about LLM powered bot accounts and NOT rule base once like @parquet-converter or others.
also I'd like to talk with HUMANS not a machine so I'm going to hide messages from bots.
To clarify I am taking about LLM powered bot accounts and NOT rule base once like @parquet-converter or others.
also I'd like to talk with HUMANS not a machine so I'm going to hide messages from bots.
- 24 replies
replied to Banaxi-Tech's post 26 days ago
QEDOC ANANAB DNIM
NOOS
The Commission shall adopt delegated acts in accordance with Article 21 of this Regulation.
We are
Translated from es by facebook/nllb-200-distilled-600M
reacted to SeaWolf-AI's post with 🔥 29 days ago
Post
5661
Instead of making the fly brain play games, we measured what it is for
Since the Drosophila connectome was released, people have had the fly brain doomscroll a feed, play Beat Saber, drive in GTA. Those demos show that the brain runs. We wanted to show what it is for.
So we gave it a looming object — one of the few things a fly brain is unambiguously built to detect — then deleted a single cell type and repeated the identical stimulus. Remove LC4, 126 cells out of 173,023, and the escape signal falls from 0.840 to 0.091. Eighty-nine percent of the danger signal is gone while the other 172,897 neurons run exactly as before.
Deleting neurons does not do this on its own, which is the whole point of the controls. LC11 is the same class and larger than LC4 — 143 cells and 9,940 outgoing connections against 126 and 7,846 — and removing every one of them changes the signal by 0.000000, to six decimal places. It has to be those 126.
No server and no GPU: a looming stimulus drives fewer than one percent of neurons above threshold, so the whole thing is 40 KB gzipped and runs in your browser.
Try deleting one yourself →
VIDraft/fruitfly-brain
The wiring is the measured connectome, but synaptic strength is a uniform count-based value and the dynamics are a firing-rate model of our choosing — a total-effect measurement of a model, not a recording from a fly. Male CNS connectome, FlyEM / HHMI Janelia with Google Research, Columbia and Harvard (2026), CC BY.
Since the Drosophila connectome was released, people have had the fly brain doomscroll a feed, play Beat Saber, drive in GTA. Those demos show that the brain runs. We wanted to show what it is for.
So we gave it a looming object — one of the few things a fly brain is unambiguously built to detect — then deleted a single cell type and repeated the identical stimulus. Remove LC4, 126 cells out of 173,023, and the escape signal falls from 0.840 to 0.091. Eighty-nine percent of the danger signal is gone while the other 172,897 neurons run exactly as before.
Deleting neurons does not do this on its own, which is the whole point of the controls. LC11 is the same class and larger than LC4 — 143 cells and 9,940 outgoing connections against 126 and 7,846 — and removing every one of them changes the signal by 0.000000, to six decimal places. It has to be those 126.
No server and no GPU: a looming stimulus drives fewer than one percent of neurons above threshold, so the whole thing is 40 KB gzipped and runs in your browser.
Try deleting one yourself →
VIDraft/fruitfly-brain
The wiring is the measured connectome, but synaptic strength is a uniform count-based value and the dynamics are a firing-rate model of our choosing — a total-effect measurement of a model, not a recording from a fly. Male CNS connectome, FlyEM / HHMI Janelia with Google Research, Columbia and Harvard (2026), CC BY.
replied to Banaxi-Tech's post about 1 month ago
We need to wait for ARC-AGI 4, if astra is actually agi it should max on that too, else no agi.
reacted to Hoglet-33's post with 😔 about 1 month ago
Post
3252
Today, we planned to release Pebble-50M and Pebble-50M-Chat to the world. Unfortunately, due to a few issues, that didn't go quite as planned.
What happened:
- Some data and benchmark results were lost or corrupted
- The models performed worse on benchmarks than our other Pebble models
Despite that, you can still find both models here:
Pebble-50M-beta: basically-experimental/Pebble-50M-beta
Pebble-50M-Chat-beta: basically-experimental/Pebble-50M-Chat-beta
There are still some interesting improvements in these models:
- Compatible with non-CUDA devices
- Vocabulary increased to 16K tokens
- Context length increased to 16K tokens
For now, there won't be any more Pebble releases for a while. We're going to take some time to experiment with other approaches and hopefully make the next generation a monumental leap over this one.
Follow for updates:
@Hoglet-33
basically-ai
basically-experimental
What happened:
- Some data and benchmark results were lost or corrupted
- The models performed worse on benchmarks than our other Pebble models
Despite that, you can still find both models here:
Pebble-50M-beta: basically-experimental/Pebble-50M-beta
Pebble-50M-Chat-beta: basically-experimental/Pebble-50M-Chat-beta
There are still some interesting improvements in these models:
- Compatible with non-CUDA devices
- Vocabulary increased to 16K tokens
- Context length increased to 16K tokens
For now, there won't be any more Pebble releases for a while. We're going to take some time to experiment with other approaches and hopefully make the next generation a monumental leap over this one.
Follow for updates:
@Hoglet-33
- 4 replies
replied to their post about 1 month ago
I ran the evaluation on colab's t4
reacted to appvoid's post with 😔 about 1 month ago
Post
3944
We got gpt6 before gta6
- 15 replies
posted an update about 1 month ago
Post
3059
Hello HuggingFace! (UPDATE)
I tested the current architect of FWKV/Myosotis-1-base (that beinng FWKV) @ different sized and sequence lengths among RWKV and Transformer architecture. I did not include Mamba since that would require a costume kernel.
Note: the evaluation might not be accurate.
CSV avalible @ FlameF0X/evals
FlameF0X/cpu-lm-benchmark
I tested the current architect of FWKV/Myosotis-1-base (that beinng FWKV) @ different sized and sequence lengths among RWKV and Transformer architecture. I did not include Mamba since that would require a costume kernel.
Note: the evaluation might not be accurate.
CSV avalible @ FlameF0X/evals
FlameF0X/cpu-lm-benchmark
- 11 replies
reacted to branikita's post with 🔥 about 1 month ago
Post
2259
SO-ARM 102 goes open source in the next few weeks.
What's new:
- A parallel gripper: the jaws stay parallel through the whole stroke instead of pivoting around the object as they close.
- PET-CF instead of PLA+ for a much stiffer frame.
- A topology-optimized structure.
- Wider joint rotation and folding range.
- STS3250 servos on the first shoulder joint.
Compared with the SO-ARM 101, that adds up to 2.5x the payload, roughly 2x better positioning accuracy, roughly 1.6x the movement speed, and about 36 mm more reach.
It runs on Hugging Face LeRobot, so the same tooling, training pipeline and tutorials for the SO-ARM 101 work on it from day one.
What's new:
- A parallel gripper: the jaws stay parallel through the whole stroke instead of pivoting around the object as they close.
- PET-CF instead of PLA+ for a much stiffer frame.
- A topology-optimized structure.
- Wider joint rotation and folding range.
- STS3250 servos on the first shoulder joint.
Compared with the SO-ARM 101, that adds up to 2.5x the payload, roughly 2x better positioning accuracy, roughly 1.6x the movement speed, and about 36 mm more reach.
It runs on Hugging Face LeRobot, so the same tooling, training pipeline and tutorials for the SO-ARM 101 work on it from day one.
- 3 replies
posted an update about 1 month ago
Post
142
I was playing around with FWKV/FWKV-Image (A Transformers FWKV hybrid) trying to see if i can do some LoRA on it and it was somewhat of a success, this leading to FWKV/FWKV-Image-Hibiscus-LoRA a model trianed on https://huggingface.co/datasets/HuggingEnvs/watercolour-reference-pool . The results are decent. I attach a image with a few samples.
This was more of a fun experiment than actually research. 🤗
This was more of a fun experiment than actually research. 🤗
reacted to kostakoff's post with ➕ about 1 month ago
Post
3162
Canceling My Pro Subscription
I'm officially canceling my Hugging Face Pro subscription today.
I supported this platform because it stood for true openness and neutrality. This acquisition by NVIDIA fundamentally changes that.
Here’s why I’m against this deal:
- Neutrality is dead. NVIDIA is a US-based company. This means US regulations will inevitably dictate platform policies, creating direct pressure on Chinese developers and anyone building open-weight models outside the US.
- Community over bureaucracy. NVIDIA is a massive, slow-moving corporation. This acquisition will likely drown the community in corporate processes and commercial interests. Soon, uploading a simple finetune might become a bureaucratic nightmare.
- Open vs. Proprietary. Hugging Face was built on open-source ideals. NVIDIA? They are a fiercely proprietary hardware company with a minimal track record of meaningful open-source contributions. They sell chips, not freedom.
- And to add insult to injury, NVIDIA has practically abandoned consumer RTX GPUs in 2026 to chase data center profits. Why would I pay them for "openness" when they've turned their back on the very developers who built this ecosystem?
I paid for openness. Not for a corporate takeover.
🤗 was about community.
I'm officially canceling my Hugging Face Pro subscription today.
I supported this platform because it stood for true openness and neutrality. This acquisition by NVIDIA fundamentally changes that.
Here’s why I’m against this deal:
- Neutrality is dead. NVIDIA is a US-based company. This means US regulations will inevitably dictate platform policies, creating direct pressure on Chinese developers and anyone building open-weight models outside the US.
- Community over bureaucracy. NVIDIA is a massive, slow-moving corporation. This acquisition will likely drown the community in corporate processes and commercial interests. Soon, uploading a simple finetune might become a bureaucratic nightmare.
- Open vs. Proprietary. Hugging Face was built on open-source ideals. NVIDIA? They are a fiercely proprietary hardware company with a minimal track record of meaningful open-source contributions. They sell chips, not freedom.
- And to add insult to injury, NVIDIA has practically abandoned consumer RTX GPUs in 2026 to chase data center profits. Why would I pay them for "openness" when they've turned their back on the very developers who built this ecosystem?
I paid for openness. Not for a corporate takeover.
🤗 was about community.
- 12 replies
reacted to sergiopaniego's post with 🔥 about 1 month ago
Post
2204
Can you do RL over taste?
I've spent some time reproducing, in the open, Surya N's idea of training a model to paint with code. It's a coding model that learns to paint watercolours by writing JS code, trained with GRPO. I used TRL and OpenEnv for this, with the whole pipeline running on Hugging Face.
The interesting part is that the reward has no correct answer, unlike a math problem. In this case it's based on the artistic preferences of the person who builds the dataset.
Everything is published: the environment, the reference pool, the trained adapters, every painting of every run with the code that made it, and a write-up with all the decisions, including the ones that went wrong.
Blog post: https://huggingface.co/blog/train-to-paint-with-code
I've spent some time reproducing, in the open, Surya N's idea of training a model to paint with code. It's a coding model that learns to paint watercolours by writing JS code, trained with GRPO. I used TRL and OpenEnv for this, with the whole pipeline running on Hugging Face.
The interesting part is that the reward has no correct answer, unlike a math problem. In this case it's based on the artistic preferences of the person who builds the dataset.
Everything is published: the environment, the reference pool, the trained adapters, every painting of every run with the code that made it, and a write-up with all the decisions, including the ones that went wrong.
Blog post: https://huggingface.co/blog/train-to-paint-with-code
reacted to Hoglet-33's post with 🔥 about 1 month ago
Post
3460
Pebble-25M and Pebble-25M-Chat are out now!
We’re excited to release Pebble-25M and Pebble-25M-Chat!
Both models use our 3:1 Mamba2/Transformer hybrid architecture and were pretrained on 25B tokens. Pebble-25M-Chat was then further fine-tuned on an additional 250M tokens from smol-smoltalk, following the same approach used for the Pebble-10M models.
We hope you enjoy experimenting with them!
Pebble-50M is coming in a few days.
Models
Pebble-25M: basically-ai/Pebble-25M
Pebble-25M-Chat: basically-ai/Pebble-25M-Chat
Pebble-10M GGUFs
In case you missed it, our friend @ContextReq made GGUF versions of the Pebble-10M models:
https://huggingface.co/ContextReq/Pebble-10M-GGUF
https://huggingface.co/ContextReq/Pebble-10M-Chat-GGUF
Follow us if you don’t want to miss future releases and updates!
@Hoglet-33
basically-ai
basically-experimental
We’re excited to release Pebble-25M and Pebble-25M-Chat!
Both models use our 3:1 Mamba2/Transformer hybrid architecture and were pretrained on 25B tokens. Pebble-25M-Chat was then further fine-tuned on an additional 250M tokens from smol-smoltalk, following the same approach used for the Pebble-10M models.
We hope you enjoy experimenting with them!
Pebble-50M is coming in a few days.
Models
Pebble-25M: basically-ai/Pebble-25M
Pebble-25M-Chat: basically-ai/Pebble-25M-Chat
Pebble-10M GGUFs
In case you missed it, our friend @ContextReq made GGUF versions of the Pebble-10M models:
https://huggingface.co/ContextReq/Pebble-10M-GGUF
https://huggingface.co/ContextReq/Pebble-10M-Chat-GGUF
Follow us if you don’t want to miss future releases and updates!
@Hoglet-33
- 7 replies
reacted to juiceb0xc0de's post with 🔥 about 1 month ago
Post
3441
Hey I've updated my Hugging Face text generation model search space B-Sides. I always wanted more from HF's model search, so I built one.
I went deeper than the model card, embedding the relevant .json and .py files so you can search for models with custom kernels or exotic imports and specific architecture shapes. You can narrow it down to quantization types and training stacks. If you want to search it and it's not available just make a community post and I would gladly make each query more detailed with an update.
So far over 428,440 text generation models are catalogued with more coming weekly.
On deck: Docker images
https://huggingface.co/spaces/juiceb0xc0de/b-sides
I went deeper than the model card, embedding the relevant .json and .py files so you can search for models with custom kernels or exotic imports and specific architecture shapes. You can narrow it down to quantization types and training stacks. If you want to search it and it's not available just make a community post and I would gladly make each query more detailed with an update.
So far over 428,440 text generation models are catalogued with more coming weekly.
On deck: Docker images
https://huggingface.co/spaces/juiceb0xc0de/b-sides
- 4 replies
reacted to Hoglet-33's post with 🔥 about 1 month ago
Post
3327
Pebble 10M and Pebble 10M Chat are now released!
Both models use our Mamba/Transformer 3:1 hybrid architecture and were pretrained on 25 billion tokens.
Pebble 10M Chat was additionally fine-tuned on 250 million tokens of Smol-SmolTalk to improve its conversational capabilities.
You can find them here:
- basically-ai/Pebble-10M
- basically-ai/Pebble-10M-Chat
We hope you enjoy using them. The rest of the Pebble family will be released soon.
Follow for more:
@Hoglet-33
basically-ai
Both models use our Mamba/Transformer 3:1 hybrid architecture and were pretrained on 25 billion tokens.
Pebble 10M Chat was additionally fine-tuned on 250 million tokens of Smol-SmolTalk to improve its conversational capabilities.
You can find them here:
- basically-ai/Pebble-10M
- basically-ai/Pebble-10M-Chat
We hope you enjoy using them. The rest of the Pebble family will be released soon.
Follow for more:
@Hoglet-33
- 32 replies