Georgi Gerganov
PulseAugur coverage of Georgi Gerganov — every cluster mentioning Georgi Gerganov across labs, papers, and developer communities, ranked by signal.
7 day(s) with sentiment data
-
llama.cpp update enhances SYCL backend for MoE scheduling
The llama.cpp project has released an update, b12251, introducing significant improvements to its SYCL backend, particularly for device-side Mixture of Experts (MoE) scheduling. This update refines how MoE experts are h…
-
llama.cpp configured for enterprise stability and multi-user access
The open-source inference engine llama.cpp, originally developed by Georgi Gerganov and now managed by ggml-org, is presented as a stable and robust alternative to engines like vLLM for enterprise environments. The guid…
-
llama.cpp b11890 release improves deepseekv4.1 integration
The llama.cpp project has released version b11890, which includes follow-up fixes and improvements specifically for the deepseekv4.1 model. Key changes involve optimizing the kv-cache using ggml_backend_tensor_copy for …
-
llama.cpp releases optimizations, new ops; contrastive learning research published
The llama.cpp project has released two pre-release versions, b11936 and b11840, introducing optimizations and new operations. Version b11936 focuses on improving the k-pool sparse selection for models like Qwen4Exp and …
-
Anthropic SDK updates, llama.cpp optimizations, and OpenAI reset theories
Recent updates to the Anthropic SDK for Python (v2.2.0) include support for message threads and enhanced usage/cost reporting. Concurrently, the llama.cpp project has released new versions (b11793 and b11774) focusing o…
-
Llama.cpp updates, Anthropic security checks, Cursor free tier, and ClaudeAI user engagement
The llama.cpp project has released version b11785, which includes optimizations for Mixture of Experts (MoE) models and support for various hardware configurations, including Apple Silicon and openEuler. Separately, Ant…
-
llama.cpp updates KV cache handling for partial state serialization
The llama.cpp project has released an update, tagged b11644, which refines how the key-value (KV) cache handles partial state serialization. This update ensures that the cache correctly honors the PARTIAL_ONLY flag, pre…
-
llama.cpp project releases multiple pre-release updates with new model support and optimizations
The llama.cpp project has released several pre-release updates, including b11559 which adds support for Gemma-4 GPU decoding and optimizes decoding for various configurations. Release b11558 addresses image limits for Q…
-
llama.cpp update b11479 enhances CUDA pool2d performance
The llama.cpp project has released an update, b11479, which includes improvements to CUDA memory reads within the pool2d function. This update also incorporates performance enhancements and the __restrict__ keyword for …
-
llama.cpp adds GLM5-Next MTP optimizations and GGUF loading
The llama.cpp project has released an update, b11474, introducing significant optimizations and features for the GLM5-Next model. This update includes the implementation of a Multi Token Prediction (MTP) graph, referred…
-
llama.cpp b11439 refactors expert copying, adds DeepSeek2 tests
The llama.cpp project has released version b11439, which includes refactoring of selective expert copying to user code. This update also incorporates tests for enrolling two models into the selective expert copy test, s…
-
llama.cpp project releases pre-release updates with performance optimizations
The llama.cpp project has released several pre-release updates, including improvements to GPU cache management for MoE experts, optimizations for various quantization types on Apple Silicon, and fixes for Metal fusion o…
-
Meta Executive Unmasked as Torrent Pirate; llama.cpp Future Discussed Post-Hugging Face Acquisition
An adult film producer has identified a prolific torrent pirate known as 'John DOE' as an executive at Meta. Separately, Georgi Gerganov, the creator of llama.cpp and GGML, has commented on the future of these projects …
-
llama.cpp's core is GGML, a C library for CPU inference
The llama.cpp project, a popular tool for running large language models on consumer hardware, is built upon a foundational C library called GGML. Created by Georgi Gerganov, GGML serves as both a lightweight tensor comp…
-
Georgi Gerganov lauded for enabling local LLM execution with llama.cpp
Georgi Gerganov is being recognized for his significant contributions to the open-source AI community through the development of llama.cpp. This project has been instrumental in making large language models more accessi…
-
AI agents to ditch cloud for local processing, says llama.cpp creator
Georgi Gerganov, the creator of llama.cpp, predicts that 90% of AI agents will move away from cloud-based infrastructure. This shift is driven by the increasing capability to run large language models, such as a 70B par…
-
Georgi Gerganov praises Qwen3.6-27B for local coding tasks
Georgi Gerganov has found the Qwen3.6-27B model to be highly effective for local coding tasks, utilizing it daily for mundane maintenance work at ggml-org. He primarily runs the model on his Apple M2 Ultra or an RTX 509…