fzliu

16 ms·

301.
A Better Way to Measure Progress in Semiconductors (2020)(spectrum.ieee.org)
1 points by fzliu 2y ago|0 comments
302.
Global AI Power Rankings(hai.stanford.edu)
1 points by fzliu 2y ago|0 comments
303.
Evaluating Code Embeddings(blog.voyageai.com)
2 points by fzliu 2y ago|0 comments
304.
Head-to-head benchmarking tool for embedding models(github.com)
2 points by fzliu 2y ago|0 comments
305.
Code embedding model with Matryoshka learning and quantization-aware training(blog.voyageai.com)
2 points by fzliu 2y ago|0 comments
306.
How do we evaluate vector-based code retrieval?(blog.voyageai.com)
1 points by fzliu 2y ago|0 comments
307.
Do LLMs Perform Latent Multi-Hop Reasoning Without Exploiting Shortcuts?(arxiv.org)
1 points by fzliu 2y ago|0 comments
308.
Multimodal RAG(github.com)
1 points by fzliu 2y ago|0 comments
309.
CS229 Lecture Notes (2023) [pdf](cs229.stanford.edu)
4 points by fzliu 2y ago|0 comments
310.
Head-to-head benchmarking tool for embedding models(github.com)
3 points by fzliu 2y ago|0 comments
311.
Embeddings Drive the Quality of RAG (2023)(blog.voyageai.com)
1 points by fzliu 2y ago|1 comments
312.
The PhD Application from a Faculty's Perspective(eugenewu.net)
3 points by fzliu 2y ago|0 comments
313.
Voyage Multilingual Rerankers(blog.voyageai.com)
1 points by fzliu 2y ago|0 comments
314.
A statistical approach to model evaluations(anthropic.com)
2 points by fzliu 2y ago|0 comments
315.
32k context length text embedding models(blog.voyageai.com)
101 points by fzliu 2y ago|32 comments
316.
Samurai: Adapting Segment Anything Model for Zero-Shot Visual Tracking(arxiv.org)
55 points by fzliu 2y ago|0 comments
317.
Embedding Model Benchmark Tool(github.com)
2 points by fzliu 2y ago|0 comments
318.
Voyage Multilingual Rerankers(blog.voyageai.com)
1 points by fzliu 2y ago|1 comments
319.
The next generation of Voyage multilingual rerankers(blog.voyageai.com)
1 points by fzliu 2y ago|0 comments
320.
by fzliu 2y ago
While the raw similarity score does matter, what typically matters more is the score relative to other documents. In the case of the examples in the notebook, those values were the highest in relative terms. I can see why this may be unclea
321.
by fzliu 2y ago
Because LLMs such as Gemini -- and other causal language models more broadly -- are trained on next token prediction, the vectors that you get from pooling the output token embeddings aren't that useful for RAG or semantic search compa
322.
by fzliu 2y ago
I understand the sentiment. We are starting to open source some tools, mostly around embedding model evaluation (i.e. https://github.com/voyage-ai/voyage-evaluation-public ), with other stuff coming up. FWIW, there are
323.
by fzliu 2y ago
No worries at all. That's great feedback and an area of improvement for us when it comes to future posts -- we'll be more explicit about multilingualism in blogs and in our docs.
324.
by fzliu 2y ago
Glad you liked the results! We do have multilingual models (and rerankers) -- voyage-3, in particular, is multilingual: https://blog.voyageai.com/2024/09/18/voyage-3/ voyage-multimodal-3 is multilingual
325.
All-in-one embedding model for interleaved text, images, and screenshots(blog.voyageai.com)
263 points by fzliu 2y ago|31 comments
326.
Multimodal embedding model for interleaved text, images, and screenshots(blog.voyageai.com)
1 points by fzliu 2y ago|0 comments
327.
BERTs Are Generative In-Context Learners(arxiv.org)
141 points by fzliu 2y ago|46 comments
328.
In Search of Forgotten Domain Generalization(arxiv.org)
1 points by fzliu 2y ago|0 comments
329.
All-in-one embedding model for interleaved text, images, and screenshots(blog.voyageai.com)
2 points by fzliu 2y ago|0 comments
330.
Age Normalized Testosterone Peaks at Series B for Male Startup Founders(arxiv.org)
2 points by fzliu 2y ago|0 comments

More ›