70% Size, 100% Accuracy: Lossless LLM Compression for Efficient GPU Inference via Dynamic-Length Float
Paper
•
2504.11651
•
Published
•
26
Julien Chaumond PRO
julien-c
AI & ML interests
<3 ML/AI for everyone, building products to propel communities fwd
Recent Activity
new activity
less than a minute ago
huggingface/HuggingDiscussions:[FEEDBACK] Inference Providers
replied to
their
post
2 days ago
Important notice 🚨
For Inference Providers who have built support for our Billing API (currently: Fal, Novita, HF-Inference – with more coming soon), we've started enabling Pay as you go (=PAYG)
What this means is that you can use those Inference Providers beyond the free included credits, and they're charged to your HF account.
You can see it on this view: any provider that does not have a "Billing disabled" badge, is PAYG-compatible.
Organizations
julien-c's activity
[FEEDBACK] Inference Providers
15
105
#49 opened 3 months ago
by
julien-c
upvoted
a
paper
20 minutes ago
replied to
their
post
2 days ago
did you get it to work since?
reacted to
their
post with 👍🚀🔥
2 days ago
Post
3644
Important notice 🚨
For Inference Providers who have built support for our Billing API (currently: Fal, Novita, HF-Inference – with more coming soon), we've started enabling Pay as you go (=PAYG)
What this means is that you can use those Inference Providers beyond the free included credits, and they're charged to your HF account.
You can see it on this view: any provider that does not have a "Billing disabled" badge, is PAYG-compatible.
For Inference Providers who have built support for our Billing API (currently: Fal, Novita, HF-Inference – with more coming soon), we've started enabling Pay as you go (=PAYG)
What this means is that you can use those Inference Providers beyond the free included credits, and they're charged to your HF account.
You can see it on this view: any provider that does not have a "Billing disabled" badge, is PAYG-compatible.
- 8 replies
[FEEDBACK] Local apps
4
23
#31 opened 11 months ago
by
kramp
Certificate of Achievement: 1. Fundamentals of LLMs
1
#1 opened 3 days ago
by
mohamed-amine-benhima
Create README.md
2
#1 opened 18 days ago
by
Gajduk
Djsmartberry
#39 opened 15 days ago
by
Djsmartberry
[ERROR] Unauthorized for Every Model
1
5
#40 opened 12 days ago
by
KingNish
Create Sporty-Video-Download,cutting app
#4 opened 16 days ago
by
joerg23
commented on
Welcome to Inference Providers on the Hub 🔥
2 days ago
in addition to all the other PRO features!
reacted to
danielhanchen's
post with ❤️🤗🔥
2 days ago
Post
3964
🦥 Introducing Unsloth Dynamic v2.0 GGUFs!
Our v2.0 quants set new benchmarks on 5-shot MMLU and KL Divergence, meaning you can now run & fine-tune quantized LLMs while preserving as much accuracy as possible.
Llama 4: unsloth/Llama-4-Scout-17B-16E-Instruct-GGUF
DeepSeek-R1: unsloth/DeepSeek-R1-GGUF-UD
Gemma 3: unsloth/gemma-3-27b-it-GGUF
We made selective layer quantization much smarter. Instead of modifying only a subset of layers, we now dynamically quantize all layers so every layer has a different bit. Now, our dynamic method can be applied to all LLM architectures, not just MoE's.
Blog with Details: https://docs.unsloth.ai/basics/dynamic-v2.0
All our future GGUF uploads will leverage Dynamic 2.0 and our hand curated 300K–1.5M token calibration dataset to improve conversational chat performance.
For accurate benchmarking, we built an evaluation framework to match the reported 5-shot MMLU scores of Llama 4 and Gemma 3. This allowed apples-to-apples comparisons between full-precision vs. Dynamic v2.0, QAT and standard iMatrix quants.
Dynamic v2.0 aims to minimize the performance gap between full-precision models and their quantized counterparts.
Our v2.0 quants set new benchmarks on 5-shot MMLU and KL Divergence, meaning you can now run & fine-tune quantized LLMs while preserving as much accuracy as possible.
Llama 4: unsloth/Llama-4-Scout-17B-16E-Instruct-GGUF
DeepSeek-R1: unsloth/DeepSeek-R1-GGUF-UD
Gemma 3: unsloth/gemma-3-27b-it-GGUF
We made selective layer quantization much smarter. Instead of modifying only a subset of layers, we now dynamically quantize all layers so every layer has a different bit. Now, our dynamic method can be applied to all LLM architectures, not just MoE's.
Blog with Details: https://docs.unsloth.ai/basics/dynamic-v2.0
All our future GGUF uploads will leverage Dynamic 2.0 and our hand curated 300K–1.5M token calibration dataset to improve conversational chat performance.
For accurate benchmarking, we built an evaluation framework to match the reported 5-shot MMLU scores of Llama 4 and Gemma 3. This allowed apples-to-apples comparisons between full-precision vs. Dynamic v2.0, QAT and standard iMatrix quants.
Dynamic v2.0 aims to minimize the performance gap between full-precision models and their quantized counterparts.
reacted to
their
post with 😎🤗🔥
2 days ago
Post
3380
BOOOOM: Today I'm dropping TINY AGENTS
the 50 lines of code Agent in Javascript 🔥
I spent the last few weeks working on this, so I hope you will like it.
I've been diving into MCP (Model Context Protocol) to understand what the hype was all about.
It is fairly simple, but still quite powerful: MCP is a standard API to expose sets of Tools that can be hooked to LLMs.
But while doing that, came my second realization:
Once you have a MCP Client, an Agent is literally just a while loop on top of it. 🤯
➡️ read it exclusively on the official HF blog: https://huggingface.co/blog/tiny-agents
the 50 lines of code Agent in Javascript 🔥
I spent the last few weeks working on this, so I hope you will like it.
I've been diving into MCP (Model Context Protocol) to understand what the hype was all about.
It is fairly simple, but still quite powerful: MCP is a standard API to expose sets of Tools that can be hooked to LLMs.
But while doing that, came my second realization:
Once you have a MCP Client, an Agent is literally just a while loop on top of it. 🤯
➡️ read it exclusively on the official HF blog: https://huggingface.co/blog/tiny-agents
- 1 reply
takedown fork
1
#9 opened 6 days ago
by
bogusred