Saiful Haq

120 posts

Saiful Haq

Saiful Haq

@RetrieveRerank

Founder (Stealth) Prev. Director of AI and Staff Research Engineer @Hyperbots_Inc IIT Bombay 5th Year CS PhD @cfiltnlp @iitbombay Building in stealth 🚀

Bengaluru, India Inscrit le Ağustos 2023
134 Abonnements99 Abonnés
Saiful Haq retweeté
Omar Khattab
Omar Khattab@lateinteraction·
overwhelming evidence for late interaction / multi-vector models yet again :-) > even after finetuning, single-vector models lag far behind multi-vector embeddings, which achieve significant performance gains and exhibit greater robustness to catastrophic forgetting.
Sumit@_reachsumit

On Strengths and Limitations of Single-Vector Embeddings Microsoft shows that dimensionality alone cannot explain poor retrieval performance of single-vector embeddings, identifying domain shift and the "drowning in documents" paradox as key factors. 📝 arxiv.org/abs/2603.29519

English
4
7
89
8.4K
Saiful Haq retweeté
Omar Khattab
Omar Khattab@lateinteraction·
The “grep-is-all-you-need” nonsense arguments arise from the fact that too many people think neural search means single-vector IR, which do in fact suck. But we’ve known that since 2019. Quoting @aaxsh18, CEO of Mixedbread: > late interaction cant stop winning
Mixedbread@mixedbreadai

For Agentic tasks, Oracle-level performance is the maximum performance a system can achieve, assuming it is able to retrieve all relevant documents perfectly, every time. We're proud to show that Mixedbread Search approaches the Oracle on multiple knowledge intensive benchmarks.

English
9
15
227
24.7K
Saiful Haq retweeté
gooby
gooby@gooby_esq·
Makes me wonder if Omar sees the rest of the world with this much clarity or just ML/AI let’s hear some hot political takes haha
Antaripa Saha@doesdatmaksense

almost everything omar and his team build looks early when it’s released, and then quietly becomes foundational later. seen the same with colbert and dspy, was not a hit immediately after release but over time the industry caught up, understood the framing, and they became widely adopted i feel it will be similar with rlms too. the industry will circle back in few months and adopt the idea of rlms as a standard way to manage computation, context, and recursion in long-running systems.

English
3
1
21
8.8K
Saiful Haq retweeté
ukituki
ukituki@ukituki·
@gooby_esq Dspy.RLM("Given all his papers reverse engineer and extract author's mental models") 😅
English
0
1
3
203
Saiful Haq retweeté
Omar Khattab
Omar Khattab@lateinteraction·
Of course not. In fact, my lab is simultaneously building RLMs as the next paradigm for LLMs *and* developing the next paradigm for retrieval (stay tuned!). Retrieval will not go anywhere: if you have a large corpus with, say, billions of tokens over which you issue many queries, you necessarily need to build some index data structures that enable fast sub-linear access. RLMs may internally choose to build such an index when it proves to be an effective tool, but fundamentally RLMs are about long one-off context. You wouldn’t typically put an RLM over a million documents and expect that to be the optimal system design. (Thank you for the question @jayitabhattac11 !)
Jayita Bhattacharyya (JB)@jayitabhattac11

Can RLMs eliminate RAG or am I hallucinating 🤔 @a1zhang @lateinteraction

English
21
17
209
16.5K
alex zhang
alex zhang@a1zhang·
For those interested in making OSS contributions to the RLM repo, I've added a bunch of random thoughts and TODOs of what to add in a *messy* Markdown file on the GH repo. Feel free to tackle any of them, or any other things you think are meaningful. I'll be pretty active here or on the repo. Once I finish some other related work, I might open up a Discord channel or something for people who want to make longer standing contributions to the repo / discuss the direction of where to take it. Cheers! github.com/alexzhang13/rl…
English
19
30
292
17.3K
Saiful Haq retweeté
Omar Khattab
Omar Khattab@lateinteraction·
@a1zhang IMO, RLMs are as “language model”-y as modern “LLMs” or Reasoning Models are truly “statistical models of language”. All three are a bit of a stretch BUT in the same way. Pedantically, all three are language processing systems, eg recursive/reasoning language processing system.
English
0
1
20
1.5K
Saiful Haq
Saiful Haq@RetrieveRerank·
@prithivida Yupp AI’s SVG leaderboard might be relevant.
English
1
0
1
80
Prithivi Da
Prithivi Da@prithivida·
Is there a benchmark that measures a vision LLM’s spatial and geometric reasoning skills ?
English
2
0
1
114
Saiful Haq retweeté
Omar Khattab
Omar Khattab@lateinteraction·
> You’ll implement ColBERT to understand multi-vector search [and] apply ColPali for patch-level image retrieval. So happy to see the great folks at @DeepLearningAI @AndrewYNg host a course on late interaction (ColBERT, ColPali et al) after their short course on DSPy :D
DeepLearning.AI@DeepLearningAI

🚀 New short course with @qdrant_engine: Multi-vector Image Retrieval. Taught by @LukawskiKacper, Senior Developer Advocate at Qdrant, the course shows how multi-vector techniques outperform single-vector methods by matching text tokens to image patches directly. You’ll implement ColBERT to understand multi-vector search, apply ColPali for patch-level image retrieval, reduce memory with quantization and pooling, and use MUVERA to enable fast HNSW search. The course concludes with a full multi-modal RAG pipeline built on ColPali and MUVERA. Learn more and enroll now: hubs.la/Q03XCQZ10

English
3
8
114
9.6K
Saiful Haq retweeté
Swaroop Nath
Swaroop Nath@swaroopnath6·
Please consider applying to the program. Over two years, my research skills, perspective on research have all been broadened and sharpened. This is an exceptional group, in the way they groom you, and allow you a room for exploring wild ideas. Pls reach out if you have questions!
Prateek Jain@jainprateek_

Thrilled to note that we are keeping the tradition of the awesome AI residency program alive in a new avatar: pre-doc researcher program at GDM-Blr -- with some amazing work done by our recent predocs including @gautham_ga_ @pranamyapk @puranjay1412 @sahilgo6801 @swaroopnath6 If you want to join this program, please apply here: google.com/about/careers/…

English
0
1
6
413
Saiful Haq retweeté
Omar Khattab
Omar Khattab@lateinteraction·
Martin @martin_casado and I had a fun hour-long chat about why we need an AI software layer, and why that's true even if AGI arrives. This is basically my take on why "the model" is definitely NOT "the product", though models are one way you may decide to implement some products
English
15
37
180
32.5K
Saiful Haq retweeté
Ankur Gupta
Ankur Gupta@getpy·
Happy Friday Everyone, DSPyWeekly Issue #11 is live! 🚀 Highlights: 🔹 A cookbook for Self-Evolving Agents 🔹 Teaching local models tool-calling 🔹 New DSPy + Neo4j integration 🔹 A new "Events" section to track DSPy meetups! Plus new projects like codex_dspy & AUTODSPy. #DSPy #AI #LLMs #AgenticAI #Neo4j
English
7
9
78
10.7K
Saiful Haq
Saiful Haq@RetrieveRerank·
@lateinteraction PEFT as an idea is clean and modular. LoRA is a bit of a hack that happens to work. Yet experimentally, it is the most effective PEFT.
English
0
0
2
528
Omar Khattab
Omar Khattab@lateinteraction·
PEFT is a great idea. I don’t know if LoRA is.
English
18
3
149
26.7K
Saiful Haq retweeté
Omar Khattab
Omar Khattab@lateinteraction·
The labs don't want you to know this (jk) but they have no clue how to best prompt their own models either. To some approximation, you just pre-/post-train it on a lot of data, intervene on certain behaviors, and what comes out is what comes out.
English
2
1
2
1K

Découvrir

@aaxsh18 @LumaHQ @gooby_esq @jayitabhattac11 @a1zhang @prithivida @MaziyarPanahi @DeepLearningAI