Re: The Era of 1-bit LLMs: ternary parameters for cost...

Fun to see ternary weights making a comeback. This was hot back in 2016 with BinaryConnect and TrueNorth chip from IBM research (disclosure, I was one of the lead chip architects there).

Authors seemed to have missed the history. They should at least cite Binary Connect or Straight Through Estimators (not my work).

Helpful hint to authors: you can get down to 0.68 bits / weight using a similar technique, good chance this will work for LLMs too.

https://arxiv.org/abs/1606.01981

This was a passion project of mine in my last few months at IBM research :).

I am convinced there is a deep connection to understanding why backprop is unreasonably effective, and the result that you can train low precision DNNs; for those note familiar, the technique is to compute the loss wrt to the low precision parameters (eg project to ternary) but apply the gradient to high precision copy of parameters (known as the straight through estimator). This is a biased estimator and there is no theoretical underpinning for why this should work, but in practice it works well.

My best guess is that it is encouraging the network to choose good underlying subnetworks to solve the problem, similar to Lottery Ticket Hypothesis. With ternary weights it is just about who connects to who (ie a graph), and not about the individual weight values anymore.

paul_mk1, 9 hours ago

#stability_ai #team #deepseek #vs #openai #comments #forecast
https://youtu.be/lY8Ja00PCQM?si=aChjauEHB0Qu_41z&t=1277

#edge #llm
https://www.androidauthority.com/openai-chatgpt-ai-device-sam-altman-3522517/

#openai #team #product
https://openai.com/index/introducing-deep-research/

Interpretable medical image Visual Question Answering via multi-modal relationship graph learning

https://www.sciencedirect.com/science/article/abs/pii/S1361841524002044

Knowledge Graphs Meet Multi-Modal Learning:
A Comprehensive Survey

https://arxiv.org/pdf/2402.05391

KAN 2.0: Kolmogorov-Arnold Networks Meet Science

https://arxiv.org/pdf/2408.10205

https://huggingface.co/papers/2502.06703

#aws #sagemaker #hyperpod #governance #cost #optimization
https://www.youtube.com/watch?v=LWOh18zSiqg

RAT: Retrieval Augmented Thoughts Elicit
Context-Aware Reasoning in Long-Horizon
Generation
https://arxiv.org/pdf/2403.05313

#LCLM #vs #RAG
In Defense of RAG in the Era of Long-Context Language Models
https://arxiv.org/pdf/2409.01666

#observability #opentelemetry #llm #traceloop #team
https://www.youtube.com/watch?v=KVgbERRPU4M

https://apxml.com/posts/gpu-requirements-deepseek-r1

#gpu #mem_req #deepseek

SCBENCH: A KV CACHE-CENTRIC ANALYSIS OF LONG-CONTEXT METHODS
https://arxiv.org/pdf/2412.10319
#MInference #LLMLingua #SnapKV #Jamba #KIVI #kvcache #benchmarks

#unsloth #team #distributed_sft #sft #fine_tuning
https://github.com/unslothai/unsloth/issues/1707#issuecomment-2658933732

https://blog.google/technology/google-deepmind/gemini-model-thinking-updates-march-2025/
#llm #new #sota #team #google #vs #openai

#MoLE #Phi #team #microsoft #multimodal
Phi-4-Mini Technical Report: Compact yet Powerful Multimodal Language Models via Mixture-of-LoRAs
https://ritvik19.medium.com/papers-explained-322-phi-4-mini-phi-4-multimodal-2be1a69be78c
https://arxiv.org/abs/2503.01743

#alibaba #team #qwen #llm
https://www.scmp.com/tech/big-tech/article/3304935/alibabas-qwen3-ai-model-coming-month-sources-say-bid-cement-industry-lead