Qwen/Qwen3.5-35B-A3B
Image-Text-to-Text • 36B • Updated
• 1M • • 993
Wang Weiyi PRO
kaupane
AI & ML interests
None yet
Recent Activity
liked
a model about 3 hours ago
Qwen/Qwen3.5-35B-A3B liked
a model 7 days ago
Qwen/Qwen3.5-27B liked
a dataset 7 days ago
Qwen/DeepPlanning Organizations
None yet
liked a model about 3 hours ago
liked a model 7 days ago
liked a dataset 7 days ago
liked a dataset 8 days ago
liked a dataset 9 days ago
updated a model 9 days ago
reacted to robtacconelli's post with 🚀 9 days ago
Post
3625
🏆 Nacrith: a 135M model that out-compresses everything on natural language
What if a tiny LM could compress english text better than _every_ compressor out there — classical or neural, small or large?
Nacrith pairs SmolLM2-135M with an ensemble of online predictors and high-precision arithmetic coding.
What's inside
The standard LLM+arithmetic coding approach wastes ~75% of CDF precision on large vocabularies. Our CDF-24 fix alone recovers 0.5 bpb. On top: a token N-gram that skips the GPU on predictable tokens, an adaptive bias head, llama.cpp backend (7× faster than PyTorch), multi-GPU parallel compression, and a binary file format (NC06) — the first LLM-based binary compressor we know of.
Runs on a GTX 1050 Ti. ~500 MB weights, ~1.2 GB VRAM per worker.
💻 Code: https://github.com/robtacconelli/Nacrith-GPU
⭐ Space: robtacconelli/Nacrith-GPU
📄 Paper: Nacrith: Neural Lossless Compression via Ensemble Context Modeling and High-Precision CDF Coding (2602.19626)
Try it, break it, share your results — all feedback welcome. ⭐ on the repo appreciated!
Results across all systems we tested:
- alice29.txt → 0.918 bpb (−44% vs CMIX, −20% vs ts_zip) — below the 2nd-order Shannon entropy bound
- enwik8 (100 MB) → 0.9389 bpb (−8% vs FineZip/LLMZip's 8B model, −15% vs ts_zip)
- Unseen text → 0.723 bpb on a doc published after training cutoff — no memorization, 26% better than FineZip/LLMZip on the same model
SmolLM2-135M by
HuggingFaceTB
What if a tiny LM could compress english text better than _every_ compressor out there — classical or neural, small or large?
Nacrith pairs SmolLM2-135M with an ensemble of online predictors and high-precision arithmetic coding.
What's inside
The standard LLM+arithmetic coding approach wastes ~75% of CDF precision on large vocabularies. Our CDF-24 fix alone recovers 0.5 bpb. On top: a token N-gram that skips the GPU on predictable tokens, an adaptive bias head, llama.cpp backend (7× faster than PyTorch), multi-GPU parallel compression, and a binary file format (NC06) — the first LLM-based binary compressor we know of.
Runs on a GTX 1050 Ti. ~500 MB weights, ~1.2 GB VRAM per worker.
💻 Code: https://github.com/robtacconelli/Nacrith-GPU
⭐ Space: robtacconelli/Nacrith-GPU
📄 Paper: Nacrith: Neural Lossless Compression via Ensemble Context Modeling and High-Precision CDF Coding (2602.19626)
Try it, break it, share your results — all feedback welcome. ⭐ on the repo appreciated!
Results across all systems we tested:
- alice29.txt → 0.918 bpb (−44% vs CMIX, −20% vs ts_zip) — below the 2nd-order Shannon entropy bound
- enwik8 (100 MB) → 0.9389 bpb (−8% vs FineZip/LLMZip's 8B model, −15% vs ts_zip)
- Unseen text → 0.723 bpb on a doc published after training cutoff — no memorization, 26% better than FineZip/LLMZip on the same model
SmolLM2-135M by
- 1 reply
upvoted an article 10 days ago
Article
Training strategies of Z-Image-Turbo
•
6
liked a Space 10 days ago
updated a Space 11 days ago
updated a dataset 12 days ago
published a Space 16 days ago
published a model 16 days ago
liked a dataset 18 days ago
liked 2 models 20 days ago
reacted to mrs83's post with 🔥 20 days ago
Post
2340
In 2017, my RNNs were babbling. Today, they are hallucinating beautifully.
10 years ago, getting an LSTM to output coherent English was a struggle.
10 years later, after a "cure" based on FineWeb-EDU and a custom synthetic mix for causal conversation, the results are fascinating.
We trained this on ~10B tokens on a single AMD GPU (ROCm). It is not a Transformer: Echo-DSRN (400M) is a novel recurrent architecture inspired by Hymba, RWKV, and xLSTM, designed to challenge the "Attention is All You Need" monopoly on the Edge.
The ambitious goal is to build a small instruct model with RAG and tool usage capabilities ( ethicalabs/Kurtis-EON1)
📊 The Benchmarks (Size: 400M)
For a model this size (trained on <10B tokens), the specialized performance is surprising:
*SciQ*: 73.8% 🦄 (This rivals billion-parameter models in pure fact retrieval).
*PIQA*: 62.3% (Solid physical intuition for a sub-1B model).
The Reality Check:
HellaSwag (29.3%) and Winogrande (50.2%) show the limits of 400M parameters and 10B tokens training.
We are hitting the "Reasoning Wall" which confirms we need to scale to (hopefully) unlock deeper common sense. As you can see in the visualization (to be released soon on HF), the FineWeb-EDU bias is strong. The model is convinced it is in a classroom ("In this course, we explore...").
The Instruct Model is not ready yet and we are currently using curriculum learning to test model plasticity.
Source code and weights will not be released yet. This is not a fork or a fine-tune: the base model is built in-house at https://www.ethicalabs.ai/, with novel components that do not exist in current open libraries.
🤝 Call for Collaboration: I am looking for Peer Reviewers interested in recurrent/hybrid architectures. If you want to explore what lies beyond Transformers, let’s connect!
Training diary: ethicalabs/Kurtis-EON1
10 years ago, getting an LSTM to output coherent English was a struggle.
10 years later, after a "cure" based on FineWeb-EDU and a custom synthetic mix for causal conversation, the results are fascinating.
We trained this on ~10B tokens on a single AMD GPU (ROCm). It is not a Transformer: Echo-DSRN (400M) is a novel recurrent architecture inspired by Hymba, RWKV, and xLSTM, designed to challenge the "Attention is All You Need" monopoly on the Edge.
The ambitious goal is to build a small instruct model with RAG and tool usage capabilities ( ethicalabs/Kurtis-EON1)
📊 The Benchmarks (Size: 400M)
For a model this size (trained on <10B tokens), the specialized performance is surprising:
*SciQ*: 73.8% 🦄 (This rivals billion-parameter models in pure fact retrieval).
*PIQA*: 62.3% (Solid physical intuition for a sub-1B model).
The Reality Check:
HellaSwag (29.3%) and Winogrande (50.2%) show the limits of 400M parameters and 10B tokens training.
We are hitting the "Reasoning Wall" which confirms we need to scale to (hopefully) unlock deeper common sense. As you can see in the visualization (to be released soon on HF), the FineWeb-EDU bias is strong. The model is convinced it is in a classroom ("In this course, we explore...").
The Instruct Model is not ready yet and we are currently using curriculum learning to test model plasticity.
Source code and weights will not be released yet. This is not a fork or a fine-tune: the base model is built in-house at https://www.ethicalabs.ai/, with novel components that do not exist in current open libraries.
🤝 Call for Collaboration: I am looking for Peer Reviewers interested in recurrent/hybrid architectures. If you want to explore what lies beyond Transformers, let’s connect!
Training diary: ethicalabs/Kurtis-EON1
- 6 replies
updated a dataset 21 days ago
liked 2 models 21 days ago
