Unfolding the Leech Lattice: Fused Multi-Shell Decoding and VRAM Layouts for 2-Bit LLM Weights
Paper โข 2609.02652 โข Published
AI & ML interests
None defined yet.
Recent Activity
View all activity
Pier-Jeanย
authored a
paper 1 day ago
Post
2762
SPECK 2 IS ALREADY OUT: specklabs/Speck2-140M
Pretrained on 4x more tokens than the previous releases (20b vs 5b).
Instruct tuned versions are coming soon.
Very interesting models are coming soon too (hint: super long context).
Thanks for everyone supporting!
Pretrained on 4x more tokens than the previous releases (20b vs 5b).
Instruct tuned versions are coming soon.
Very interesting models are coming soon too (hint: super long context).
Thanks for everyone supporting!
- 2 replies
[Support] Community Articles
๐๐ฅ 2
113
#5 opened over 2 years ago
by 
victor
Join request pending for a week, anything else needed?
1
#28 opened about 1 month ago
by 
yuncheol-ncdtech
[Support] Community Articles
๐๐ฅ 2
113
#5 opened over 2 years ago
by victor
Post
138
NEW SPECK UPDATES:
Just hit #14 and #15 with out FIRST models on Open SLM Leaderboard. The models were trained on 5B tokens, while competing with similarly sized models trained on more than 6-20x the data.
A new base model Speck1.5-140M being trained right now on a higher quality corpus and will be released soon.
SpeckChat3 is coming very soon with 1 million samples, specifically designed to post train small base models.
Also, just to clarify stuff, we will NOT release anything that is NOT MIT licensed EVER. Openness is needed in small language research.
Thanks to everyone supporting the project, and stay tuned for new releases!
Just hit #14 and #15 with out FIRST models on Open SLM Leaderboard. The models were trained on 5B tokens, while competing with similarly sized models trained on more than 6-20x the data.
A new base model Speck1.5-140M being trained right now on a higher quality corpus and will be released soon.
SpeckChat3 is coming very soon with 1 million samples, specifically designed to post train small base models.
Also, just to clarify stuff, we will NOT release anything that is NOT MIT licensed EVER. Openness is needed in small language research.
Thanks to everyone supporting the project, and stay tuned for new releases!
Post
1876
SPECK UPDATES:
1 New instruct model tuned on top of Speck1-140M: specklabs/Speck1-140M-Instruct
2 Instruction tuning datasets
2 GGUFs
Much more coming soon:
Speck1.1-140M-Instruct that is post trained on SpeckChat2 will be coming very soon
New base model Speck1.5-140M is coming with a much higher quality corpus
Thanks to everyone who is already supporting the project, and stay tuned for new releases!
1 New instruct model tuned on top of Speck1-140M: specklabs/Speck1-140M-Instruct
2 Instruction tuning datasets
2 GGUFs
Much more coming soon:
Speck1.1-140M-Instruct that is post trained on SpeckChat2 will be coming very soon
New base model Speck1.5-140M is coming with a much higher quality corpus
Thanks to everyone who is already supporting the project, and stay tuned for new releases!
- 3 replies
Post
2116
FIRST SPECK MODEL RELEASED:
specklabs/Speck1-140M
new models coming very soon (both instruct and much better models), with much much higher training scale as i am getting marenostrum5 access soon!
we will be looking at 100b-2t token budgets :)
specklabs/Speck1-140M
new models coming very soon (both instruct and much better models), with much much higher training scale as i am getting marenostrum5 access soon!
we will be looking at 100b-2t token budgets :)
- 4 replies
Post
2064
Built a CPU-only BPM + musical key detector with Essentia and Gradio.
It returns BPM, key/scale, Camelot code, confidence, and analyzed duration. A representative 90-second window keeps CPU latency bounded; half-time/double-time and key-changing tracks remain the main edge cases.
Implementation notes and code:
https://huggingface.co/blog/Gboo2023/detect-bpm-musical-key-cpu-essentia-gradio
#audio #music-information-retrieval #Essentia #Gradio
Gboo2023/vocaldock-bpm-key-finder
It returns BPM, key/scale, Camelot code, confidence, and analyzed duration. A representative 90-second window keeps CPU latency bounded; half-time/double-time and key-changing tracks remain the main edge cases.
Implementation notes and code:
https://huggingface.co/blog/Gboo2023/detect-bpm-musical-key-cpu-essentia-gradio
#audio #music-information-retrieval #Essentia #Gradio
Gboo2023/vocaldock-bpm-key-finder
[Support] Community Articles
๐๐ค 2
113
#5 opened over 2 years ago
by victor
[Support] Community Articles
๐๐ฅ 2
113
#5 opened over 2 years ago
by victor
kingabzproย
posted an update about 2 months ago
Post
808
How does billing work for Hugging Face Inference Providers? If I already have Hugging Face GPU credits, can I use them, or do I need to add a credit card and pay separately?
- 3 replies
scthorntonย
posted an update about 2 months ago
Post
2682
SecureCode update: we went back and fact-checked our own security dataset and corrected what didn't hold up.
The original claim was "complete incident grounding, every example ties to a documented CVE." An adversarial re-audit found that it was overstated: many CVEs were misattributed, and many "incidents" were representative scenarios carrying invented statistics. So we fixed it.
- Grounding: re-verified every reference. Removed 802 misattributed CVEs on the web side, corrected or honestly relabeled the incident narratives, and confirmed the AI/ML conversation CVEs are real (EchoLeak CVE-2025-32711, EmailGPT CVE-2024-5184, and others).
- Fix-correctness: reviewed whether each "secure" example actually eliminates the vulnerability. Removed 28 that did not (a "secure" secret scanner whose entropy check always returned zero, an Angular example still using bypassSecurityTrustHtml, and more).
- Leakage: re-split so near-duplicates stay on one side. Test contamination went from 11.6% to zero.
- Viewer, schema, and metadata: rebuilt as parquet under a shared schema. All three viewers are live.
- Models: retrained the whole family on the corrected data so the fix reaches the weights, not just the cards. Now ten open models (3B to 26B), including two new Gemma 4 variants, refreshed locally on a DGX Spark GB10. The paper (arXiv:2512.18542) was revised to match.
Counts moved from 2,185 to 2,372 unified (web 1,625 + AI/ML 747). A slightly smaller, fully-checked dataset beats a larger one you have to take on faith. Full writeup and links in the article.
Datasets: scthornton/securecode, scthornton/securecode-web, scthornton/securecode-aiml
The original claim was "complete incident grounding, every example ties to a documented CVE." An adversarial re-audit found that it was overstated: many CVEs were misattributed, and many "incidents" were representative scenarios carrying invented statistics. So we fixed it.
- Grounding: re-verified every reference. Removed 802 misattributed CVEs on the web side, corrected or honestly relabeled the incident narratives, and confirmed the AI/ML conversation CVEs are real (EchoLeak CVE-2025-32711, EmailGPT CVE-2024-5184, and others).
- Fix-correctness: reviewed whether each "secure" example actually eliminates the vulnerability. Removed 28 that did not (a "secure" secret scanner whose entropy check always returned zero, an Angular example still using bypassSecurityTrustHtml, and more).
- Leakage: re-split so near-duplicates stay on one side. Test contamination went from 11.6% to zero.
- Viewer, schema, and metadata: rebuilt as parquet under a shared schema. All three viewers are live.
- Models: retrained the whole family on the corrected data so the fix reaches the weights, not just the cards. Now ten open models (3B to 26B), including two new Gemma 4 variants, refreshed locally on a DGX Spark GB10. The paper (arXiv:2512.18542) was revised to match.
Counts moved from 2,185 to 2,372 unified (web 1,625 + AI/ML 747). A slightly smaller, fully-checked dataset beats a larger one you have to take on faith. Full writeup and links in the article.
Datasets: scthornton/securecode, scthornton/securecode-web, scthornton/securecode-aiml
- 8 replies
cfahlgren1ย
submitted a
paper to Daily Papers 3 months ago
Pending Access Request to Join HF Blog Explorers
6
#17 opened 5 months ago
by 
AINovice2005
merveย
updated a
Space 4 months ago
[Support] Community Articles
๐๐ฅ 2
113
#5 opened over 2 years ago
by victor
[Support] Community Articles
๐๐ฅ 2
113
#5 opened over 2 years ago
by victor
CRAFTFrameworkย
posted an update 4 months ago
Post
117
# A measurable QA layer for LLM working sessions
Hallucination is treated as an inherent LLM failure mode, but most production workflows respond to it with "be careful, double-check things." That doesn't scale past a handful of sessions, and it doesn't catch the failure mode that does the most damage: confident reconstruction in late-session context.
CRAFT for Cowork takes a structural approach. The QA framework runs verification at four levels โ individual claims, recipe execution, file integrity, and cross-session consistency โ and treats trust as a measurable property rather than a vibe.
**The four-gate verification sub-routine** (RCP-CWK-024) runs before any recipe reports a result:
1. *File-pointability* โ claim traceable to a specific file
2. *Read-vs-reconstructed* โ was data actually read this session
3. *Lessons-Learned conflict* โ contradicts documented prior truth
4. *Untested assumption* โ verified vs. assumed
**Confidence scoring** grades every factual claim 0-100 against a source hierarchy: evidence read from files (80-100), tool-output observation (50-79), design intent (30-49), pure reasoning (0-29). A 10-point penalty applies past 70% token usage to correct for late-session reliability decay.
**Cross-session consistency** is enforced by a longitudinal audit recipe (RCP-CWK-036) run every 5-10 sessions. It has caught ~40% drift in tracking-file state tables โ drift that would otherwise propagate as silent ground truth.
**Concrete result:** A factual claim validation pass caught nine pre-publication content files referencing the framework with an incorrect license descriptor. Single pass, all nine corrected.
This is week 5 of an 8-week capability spotlight. CRAFT for Cowork is a free public beta.
Repository: https://github.com/CRAFTFramework/craft-framework
License: https://craftframework.ai/craft-license/ (Spec under BSL 1.1, converts to Apache 2.0 on Jan 1, 2029; content proprietary)
Hallucination is treated as an inherent LLM failure mode, but most production workflows respond to it with "be careful, double-check things." That doesn't scale past a handful of sessions, and it doesn't catch the failure mode that does the most damage: confident reconstruction in late-session context.
CRAFT for Cowork takes a structural approach. The QA framework runs verification at four levels โ individual claims, recipe execution, file integrity, and cross-session consistency โ and treats trust as a measurable property rather than a vibe.
**The four-gate verification sub-routine** (RCP-CWK-024) runs before any recipe reports a result:
1. *File-pointability* โ claim traceable to a specific file
2. *Read-vs-reconstructed* โ was data actually read this session
3. *Lessons-Learned conflict* โ contradicts documented prior truth
4. *Untested assumption* โ verified vs. assumed
**Confidence scoring** grades every factual claim 0-100 against a source hierarchy: evidence read from files (80-100), tool-output observation (50-79), design intent (30-49), pure reasoning (0-29). A 10-point penalty applies past 70% token usage to correct for late-session reliability decay.
**Cross-session consistency** is enforced by a longitudinal audit recipe (RCP-CWK-036) run every 5-10 sessions. It has caught ~40% drift in tracking-file state tables โ drift that would otherwise propagate as silent ground truth.
**Concrete result:** A factual claim validation pass caught nine pre-publication content files referencing the framework with an incorrect license descriptor. Single pass, all nine corrected.
This is week 5 of an 8-week capability spotlight. CRAFT for Cowork is a free public beta.
Repository: https://github.com/CRAFTFramework/craft-framework
License: https://craftframework.ai/craft-license/ (Spec under BSL 1.1, converts to Apache 2.0 on Jan 1, 2029; content proprietary)
- 1 reply