If
Labif
·
AI & ML interests
None yet
Recent Activity
liked a dataset 8 days ago
nvidia/simready-assets reacted to SeaWolf-AI's post with 🚀 8 days ago
No GPU. No cloud. Just your phone.
Most “small” models still beg for a datacenter. Ours fit in your pocket.
VIDRAFT POCKET is a family of GGUF models that run on an iPhone and on a GPU-free PC, straight out of the box with llama.cpp, Ollama, LM Studio, and PocketPal. Seven weeks after launch they crossed 1 million downloads. Today the family is past 2.81 million and still climbing, every day.
Three models, one pocket:
🟢 POCKET-35B — a 35B mixture-of-experts that runs on an iPhone and on a PC with no GPU. Balanced power.
→ https://huggingface.co/FINAL-Bench/POCKET-35B-GGUF
🔵 POCKET-26B — a Gemma-4 based 26B MoE, tuned for on-device Korean. Lightweight and fast.
→ https://huggingface.co/FINAL-Bench/POCKET-26B-GGUF
🟣 POCKET-Darwin-180B — a 180B flagship running on a laptop in 4-bit GGUF. The datacenter, on your desk.
→ https://huggingface.co/FINAL-Bench/POCKET-Darwin-180B-GGUF
The whole family lives here: https://huggingface.co/FINAL-Bench
Try our models live in the browser: https://huggingface.co/spaces/FINAL-Bench/POCKET-Image-Studio
Private by default. Offline by design. Yours to run.
#OnDeviceAI #LocalLLM #llamacpp #GGUF #VIDRAFT #POCKET reacted to tardellirs's post with 🔥 8 days ago
Model cards say what a model is: a fine-tune of X, a quantization of Y. I built Weightprint to check that against the weights.
App: https://huggingface.co/spaces/tardellirs/weightprint
Data (every pair, and the copy groups): https://huggingface.co/datasets/tardellirs/weightprint
It reads the header and a few bytes of every tensor (no downloads) and compares a model with the parent its card declares. 120,976 models so far. Some of what came out:
- 7% of the declared fine-tunes we could compare (1,630 of 22,847) hold their parent's weights unchanged. (Corrected from 12%: quants whose values the engine doesn't decode, like bitsandbytes and FP8, had been counted as unchanged. Thanks @dipankarsarkar for the question that caught it.)
- 27% of the fine-tunes we could compare are LoRA-style: only the attention and MLP matrices moved, norms and embeddings untouched. That share went from 10% in 2023 to 28% in 2026.
- Of 1,668 fine-tunes named abliterated, uncensored or heretic, 53% edit only the matrices that write to the residual stream, the signature of an abliteration. The rest were trained further, merged, or are the parent's weights.
- 51% of the weight bytes on the Hub are byte-for-byte repeats of another repo's files.
Search any Hub id to see which layers and parts changed against its parent, or compare any two models.
A sample per tensor can miss things, and some formats can't be compared yet. If a verdict on a model you know looks wrong, tell me here and I'll check it against the engine. That's what this post is for, before a longer write-up.Organizations
None yet