LightNews LightNews
serpwise.bsky.social
@serpwise.bsky.social
15 followers 310 following 0 posts
Posts Replies Media Videos
Reposted
nic221.bsky.social
Beyond RAG: SEARCH-R1 integrates search engines directly into reasoning models https://venturebeat.com/ai/beyond-rag-search-r1-integrates-search-engines-directly-into-reasoning-models/ #AI #search #reasoning
Text Shot: SEARCH-R1 enables LLMs to interact with search engines during their reasoning process as opposed to having a separate retrieval stage. SEARCH-R1 defines the search engine as part of the LLM’s environment, enabling the model to integrate its token generation with search engine results seamlessly.  The researchers designed SEARCH-R1 to support iterative reasoning and search. The model is trained to generate separate sets of tokens for thinking, search, information, and answer segments
March 19, 2025 at 3:48 PM
Reposted
axsaucedo.bsky.social
For production ML practitioners it is often important to develop best practices particularly when contributing to large codebases - these are the most common mistakes to avoid:

Article: www.seangoedecke.com/large-establ...

#ML #MachineLearning #ArtificialIntelligence #AI #MLOps #AIOps
March 18, 2025 at 6:03 PM
Reposted
amyhoy.bsky.social
found a really neat tool that automatically renames screenshots & other files based on their contents using an LLM

a good use case. not infringing.

alas it does NOT work with a model you can run locally, efficiently, one that won't be full of stolen data
March 19, 2025 at 7:05 PM
Reposted
tearche.bsky.social
AI agents are coming: manus.im.

Innocent enough now but in time, a great danger.

#AIAgent #Manus #AI
Manus
Manus is a general AI agent that turns your thoughts into actions. It excels at various tasks in work and life, getting everything done while you rest.
manus.im
March 7, 2025 at 12:11 AM
Reposted
sungkim.bsky.social
Stability AI's Stable Virtual Camera: Generative View Synthesis with Diffusion Models (Not Commercial License)

Given any number of input views and their cameras, it generates novel views of a scene at any target camera of interest.
March 19, 2025 at 3:01 AM
Reposted
sungkim.bsky.social
Nvidia's Llama-Nemotron models. These models allow you to set reasoning ON/OFF during runtime.

They also released all the post-training data under CC-BY-4!

Blog: developer.nvidia.com/blog/build-e...
Models and Dataset: huggingface.co/collections/...
Demo: build.nvidia.com/nvidia/llama...
Build Enterprise AI Agents with Advanced Open NVIDIA Llama Nemotron Reasoning Models | NVIDIA Technical Blog
Organizations are embracing AI agents to enhance productivity and streamline operations. To maximize their impact, these agents need strong reasoning abilities to navigate complex problems…
developer.nvidia.com
March 19, 2025 at 3:07 AM
Reposted
sungkim.bsky.social
They finally released the turbo models and FlashVDM repo.

- Hunyuan3D-2mv ( huggingface.co/tencent/Huny... )
- Hunyuan3D-2mv Demo ( huggingface.co/spaces/tence... )
- Hunyuan3D-2mini ( huggingface.co/tencent/Huny... )
- Hunyuan3D-2mini Demo ( huggingface.co/spaces/tence... )
sungkim.bsky.social Sung Kim @sungkim.bsky.social · Mar 19
Tencent (i.e., the biggest gaming company) releasing Hunyuan3D-Turbo in a hour or so.
March 19, 2025 at 4:25 PM
Reposted
adinayakup.bsky.social
Step-Video-TI2V 🔥 text driven Image-to-Video Model released by StepFun

Paper:
huggingface.co/papers/2503....
Model:
huggingface.co/stepfun-ai/s...

✨ 30B with MIT license
✨ Up to 102 frames from text + image inputs
✨ Supports motion control and anime-style
Paper page - Step-Video-TI2V Technical Report: A State-of-the-Art Text-Driven Image-to-Video Generation Model
Join the discussion on this paper page
huggingface.co
March 19, 2025 at 3:35 PM