Ganesh
@ganesh3.bsky.social
Machine Learning with Patents granted & filed | Current focus on LLM/GenAI | Graph Analytics
#Memento - new framework lets #LLM #agents learn from experience, no fine-tuning required.
The system has 3 main components: a #planner, a #tool-enabled executor that work in an alternating loop to complete tasks, & a growing "case bank" that stores #past #experiences
venturebeat.com/ai/this-new-...
The system has 3 main components: a #planner, a #tool-enabled executor that work in an alternating loop to complete tasks, & a growing "case bank" that stores #past #experiences
venturebeat.com/ai/this-new-...
venturebeat.com
September 15, 2025 at 11:04 PM
#Memento - new framework lets #LLM #agents learn from experience, no fine-tuning required.
The system has 3 main components: a #planner, a #tool-enabled executor that work in an alternating loop to complete tasks, & a growing "case bank" that stores #past #experiences
venturebeat.com/ai/this-new-...
The system has 3 main components: a #planner, a #tool-enabled executor that work in an alternating loop to complete tasks, & a growing "case bank" that stores #past #experiences
venturebeat.com/ai/this-new-...
Making #Kernel Development more accessible with #KernelLLM
We introduce KernelLLM, a large language model #LLM based on #Llama 3.1 Instruct, which has been trained specifically for the task of authoring #GPU kernels using #Triton
huggingface.co/facebook/Ker...
We introduce KernelLLM, a large language model #LLM based on #Llama 3.1 Instruct, which has been trained specifically for the task of authoring #GPU kernels using #Triton
huggingface.co/facebook/Ker...
facebook/KernelLLM · Hugging Face
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
huggingface.co
May 20, 2025 at 10:50 PM
Making #Kernel Development more accessible with #KernelLLM
We introduce KernelLLM, a large language model #LLM based on #Llama 3.1 Instruct, which has been trained specifically for the task of authoring #GPU kernels using #Triton
huggingface.co/facebook/Ker...
We introduce KernelLLM, a large language model #LLM based on #Llama 3.1 Instruct, which has been trained specifically for the task of authoring #GPU kernels using #Triton
huggingface.co/facebook/Ker...
#miniCOIL: on the Road to Usable #Sparse #Neural Retrieval. Use a vector that is comprised of contextual meaning of the words from a dense model and if word is not seen during trying then fall back on #BM25 which leverages the TF-IDF based approach.
qdrant.tech/articles/min...
qdrant.tech/articles/min...
miniCOIL: on the Road to Usable Sparse Neural Retrieval - Qdrant
Introducing miniCOIL, a lightweight sparse neural retriever capable of generalization.
qdrant.tech
May 19, 2025 at 4:20 AM
#miniCOIL: on the Road to Usable #Sparse #Neural Retrieval. Use a vector that is comprised of contextual meaning of the words from a dense model and if word is not seen during trying then fall back on #BM25 which leverages the TF-IDF based approach.
qdrant.tech/articles/min...
qdrant.tech/articles/min...
#AlphaEvolve: A #Gemini-powered #coding agent for designing #advanced #algorithms. New #AI agent evolves algorithms for #math and practical applications in computing by combining the creativity of large language models #LLMs with automated #evaluators
deepmind.google/discover/blo...
deepmind.google/discover/blo...
AlphaEvolve: A Gemini-powered coding agent for designing advanced algorithms
New AI agent evolves algorithms for math and practical applications in computing by combining the creativity of large language models with automated evaluators
deepmind.google
May 18, 2025 at 8:15 AM
#AlphaEvolve: A #Gemini-powered #coding agent for designing #advanced #algorithms. New #AI agent evolves algorithms for #math and practical applications in computing by combining the creativity of large language models #LLMs with automated #evaluators
deepmind.google/discover/blo...
deepmind.google/discover/blo...
Build and #train a #recommender system in 10 minutes using #Keras and #JAX or #tensorflow or #pytorch
developers.googleblog.com/en/build-tra...
developers.googleblog.com/en/build-tra...
Build and train a recommender system in 10 minutes using Keras and JAX- Google Developers Blog
Explore KerasRS, a new library built on Keras 3 with state-of-the-art recommendation techniques with multi-backend support.
developers.googleblog.com
May 18, 2025 at 8:11 AM
Build and #train a #recommender system in 10 minutes using #Keras and #JAX or #tensorflow or #pytorch
developers.googleblog.com/en/build-tra...
developers.googleblog.com/en/build-tra...
An extremely fast #Python #type #checker and #language #server, written in #Rust
github.com/astral-sh/ty
github.com/astral-sh/ty
GitHub - astral-sh/ty: An extremely fast Python type checker and language server, written in Rust.
An extremely fast Python type checker and language server, written in Rust. - astral-sh/ty
github.com
May 18, 2025 at 8:08 AM
#Orbital converts trained #scikit-#learn #pipelines into pure #SQL, enabling #machinelearning model execution directly within #databases—no #Python runtime needed.
posit-dev.github.io/orbital/
posit-dev.github.io/orbital/
Orbital
posit-dev.github.io
May 13, 2025 at 10:31 PM
#Orbital converts trained #scikit-#learn #pipelines into pure #SQL, enabling #machinelearning model execution directly within #databases—no #Python runtime needed.
posit-dev.github.io/orbital/
posit-dev.github.io/orbital/
Use #Amazon #Bedrock #Intelligent #Prompt #Routing within #LLM model family for #cost and #latency benefits aws.amazon.com/blogs/machin...
Use Amazon Bedrock Intelligent Prompt Routing for cost and latency benefits | Amazon Web Services
Today, we’re happy to announce the general availability of Amazon Bedrock Intelligent Prompt Routing. In this blog post, we detail various highlights from our internal testing, how you can get started...
aws.amazon.com
May 4, 2025 at 10:52 PM
Use #Amazon #Bedrock #Intelligent #Prompt #Routing within #LLM model family for #cost and #latency benefits aws.amazon.com/blogs/machin...
Flexible, lightweight #opensource #framework for #orchestrating #multiple #AI #agents to handle #complex conversations. Features include 🧠 Intelligent #intent #classification, 🔤 Dual #language support, 🌊 Flexible agent responses, 📚 #Context management
github.com/awslabs/mult...
github.com/awslabs/mult...
GitHub - awslabs/multi-agent-orchestrator: Flexible and powerful framework for managing multiple AI agents and handling complex conversations
Flexible and powerful framework for managing multiple AI agents and handling complex conversations - awslabs/multi-agent-orchestrator
github.com
May 4, 2025 at 10:33 PM
Flexible, lightweight #opensource #framework for #orchestrating #multiple #AI #agents to handle #complex conversations. Features include 🧠 Intelligent #intent #classification, 🔤 Dual #language support, 🌊 Flexible agent responses, 📚 #Context management
github.com/awslabs/mult...
github.com/awslabs/mult...
An #opensource #tool for seamless migration from other #LLMs to #Llama, and for general #prompt #optimization
github.com/meta-llama/l...
github.com/meta-llama/l...
GitHub - meta-llama/llama-prompt-ops: An open-source tool for seamless migration from other LLMs to Llama, and for general prompt optimization.
An open-source tool for seamless migration from other LLMs to Llama, and for general prompt optimization. - meta-llama/llama-prompt-ops
github.com
May 4, 2025 at 10:28 PM
An #opensource #tool for seamless migration from other #LLMs to #Llama, and for general #prompt #optimization
github.com/meta-llama/l...
github.com/meta-llama/l...
Introducing #HallOumi:
A State-of-the-Art Claim-Verification Model - a family of #opensource claim verification (hallucination detection) models, outperforming #DeepSeek R1, #OpenAI o1, #Google Gemini 1.5 Pro, #Llama 3.1 405B, and #Claude Sonnet 3.5 at only 8B parameters!
oumi.ai/blog/posts/i...
A State-of-the-Art Claim-Verification Model - a family of #opensource claim verification (hallucination detection) models, outperforming #DeepSeek R1, #OpenAI o1, #Google Gemini 1.5 Pro, #Llama 3.1 405B, and #Claude Sonnet 3.5 at only 8B parameters!
oumi.ai/blog/posts/i...
Oumi - Introducing HallOumi, A State-of-the-Art Claim-Verification Model
We're excited to introduce HallOumi-8B and HallOumi-8B-Classifier, a family of open-source claim verification (hallucination detection) models, outperforming DeepSeek R1, OpenAI o1, Google Gemini 1.5 ...
oumi.ai
April 7, 2025 at 8:03 AM
Introducing #HallOumi:
A State-of-the-Art Claim-Verification Model - a family of #opensource claim verification (hallucination detection) models, outperforming #DeepSeek R1, #OpenAI o1, #Google Gemini 1.5 Pro, #Llama 3.1 405B, and #Claude Sonnet 3.5 at only 8B parameters!
oumi.ai/blog/posts/i...
A State-of-the-Art Claim-Verification Model - a family of #opensource claim verification (hallucination detection) models, outperforming #DeepSeek R1, #OpenAI o1, #Google Gemini 1.5 Pro, #Llama 3.1 405B, and #Claude Sonnet 3.5 at only 8B parameters!
oumi.ai/blog/posts/i...
A #Visual Guide to Mixture of Experts (#MoE). Demystifying the role of MoE in Large Language Models #LLM
newsletter.maartengrootendorst.com/p/a-visual-g...
newsletter.maartengrootendorst.com/p/a-visual-g...
A Visual Guide to Mixture of Experts (MoE)
Demystifying the role of MoE in Large Language Models
newsletter.maartengrootendorst.com
April 7, 2025 at 6:54 AM
A #Visual Guide to Mixture of Experts (#MoE). Demystifying the role of MoE in Large Language Models #LLM
newsletter.maartengrootendorst.com/p/a-visual-g...
newsletter.maartengrootendorst.com/p/a-visual-g...
Do we really have to employ #complex autonomous #software #agents? #Agentless -- an agentless approach employs a simplistic three-phase process of #localization, #repair, and #patch validation, without letting the #LLM decide future actions #opensource #software #agents
arxiv.org/abs/2407.01489
arxiv.org/abs/2407.01489
Agentless: Demystifying LLM-based Software Engineering Agents
Recent advancements in large language models (LLMs) have significantly advanced the automation of software development tasks, including code synthesis, program repair, and test generation. More recent...
arxiv.org
January 14, 2025 at 10:06 PM
Do we really have to employ #complex autonomous #software #agents? #Agentless -- an agentless approach employs a simplistic three-phase process of #localization, #repair, and #patch validation, without letting the #LLM decide future actions #opensource #software #agents
arxiv.org/abs/2407.01489
arxiv.org/abs/2407.01489
#Sky-T1: Train your own #O1 preview #model within $450 #opensource #LLM #reasoning
novasky-ai.github.io/posts/sky-t1/
novasky-ai.github.io/posts/sky-t1/
Sky-T1: Train your own O1 preview model within $450
We introduce Sky-T1-32B-Preview, our reasoning model that performs on par with o1-preview on popular reasoning and coding benchmarks.
novasky-ai.github.io
January 14, 2025 at 8:49 AM
#Sky-T1: Train your own #O1 preview #model within $450 #opensource #LLM #reasoning
novasky-ai.github.io/posts/sky-t1/
novasky-ai.github.io/posts/sky-t1/
