immortal3

4 ms·

1.
Improving LLM Inference with Continuous Batching: Orca Through Tinyorca(junupark.xyz)
2 points by immortal3 6mo ago|0 comments
2.
Bits-per-Byte (BPB): a tokenizer-agnostic way to measure LLMs(dipkumar.dev)
1 points by immortal3 1y ago|0 comments
3.
by immortal3 1y ago
There's another angle to this comparison. Groq and Cerebras use custom chips, but I'm not sure about Together. In this case, Together is sharing results based on the B200 GPU. Another important point is the accuracy of these speed
4.
Creativity Is a Luxury(dipkumar.dev)
2 points by immortal3 1y ago|0 comments
5.
GPT-5 Router – Inevitable Future of Chat Interfaces(dipkumar.dev)
3 points by immortal3 1y ago|0 comments
6.
Instruction Aware Embeddings – Why Your Retriever Is Failing(dipkumar.dev)
1 points by immortal3 1y ago|0 comments
7.
Improving Retrieval in RAG (Via Recall, Precision, and NDCG)(dipkumar.dev)
2 points by immortal3 2y ago|0 comments
8.
by immortal3 2y ago
Is it possible that US or any other government has participated in the round which might be the reason?
9.
by immortal3 2y ago
Honestly, it doesn't matter for the end user if there are more tokens generated between the AI reply and human message. This is like getting rid of AI wrappers for specific tasks. If the jump in accuracy is actual, then for all practic
10.
Show HN:AceVocab - Learn and master the vocabulary featured in the GRE/GMAT(acevocab.com)
3 points by immortal3 2y ago|0 comments
11.
by immortal3 2y ago
After the acquisition, I think Wiz would have to only focus on Google Cloud which might be a major limiting factor in the company's future. But other than that, It surprises me that, a $23B offer is rejected from the perspective of Emp
12.
AWS BedRock – Converse API – A single endpoint for all models?(dipkumar.dev)
2 points by immortal3 2y ago|0 comments
13.
Essential Database Design: Five Fields Every Table Must Have(dipkumar.dev)
3 points by immortal3 2y ago|1 comments
14.
India issues notice to Google for blocking count over nude childhood photo(deccanherald.com)
3 points by immortal3 3y ago|0 comments
15.
by immortal3 3y ago
To add further, i do not know what is end goal of hugging face. 1. They have inference API but all cloud provider can implement those in next year. 2. They offer subscription but market-size of subscription is questionable. 3. I hope this
16.
Hugging Face raises $235M from investors including Salesforce and Nvidia(techcrunch.com)
378 points by immortal3 3y ago|203 comments
17.
Speeding up the GPT with KV cache (memoization)(immortal3.github.io)
2 points by immortal3 4y ago|0 comments
18.
by immortal3 5y ago
You can view ipython notebook easily with nbviewer. (sometime github rendering fails for large notebooks) This year AOC 2021: https://nbviewer.org/github/norvig/pytudes/blob/main/ipynb/A...
19.
by immortal3 5y ago
I have been thinking about these a lot lately. There is lot more scope on improving current search results. I believe simple two stage system might be enough to produce decent system. Stage - 1. query expansion and reverse index based retri