Açık kaynak radarı

Havuz

Takip ettiğimiz konularda yükselen açık kaynak projeler — yıldız ivmesi, gerçek geliştirme aktivitesi ve AI değerlendirmesiyle.

103

repo · sayfa 1 / 3

Sırala

Yıldız/gün (hız)Son gün (★/gün)Son hafta (★/gün)Son ay (★/gün)Toplam yıldız AI önem skoru Bana uyum Commit (28g)Yazar (28g)En yeni

Konu

Tümü AI / LLM18 Veri9 Geliştirici araçları13 Diğer3 SaaS altyapı13 Web11

AI / LLM

Tümü ai-agents 5.534 llm 4.770 mcp 4.267 claude 2.765 anthropic 1.242 openai 939 rag 853 agents 787 prompt-engineering 769 image-generation 545 knowledge-graph 374 generative-ai 358 text-to-speech 293 semantic-search 213 speech-recognition 164 embeddings 153 vector-search 137 llmops 108

Süz

Yalnız hareketli Yalnız analizli≥ 0 ≥ 100 ≥ 500 ≥ 2.000

Yalnız hareketli, son 28 günde commit almamış repoları gizler (yıldızı olan ama bırakılmış projeler). Commit sayısı henüz ölçülmemiş repolar da gizlenir.

EtiketTüm GitHub etiketleri — repo sahiplerinin koyduğu topic'lergöster

Sırala:

Etiketler yükleniyor…

Aktif quantization

1
FareedKhan-dev/kimi-k3-in-c önem 38gpt6-luna uyum 20gpt6-luna 9.465
kimi-k3-in-c, Kimi K3 modelinde inference çalıştırmayı hedefleyen bağımlılıksız bir C99 motorudur.
Neden yükseliyor · 57 günde 8.683 yıldız alması ve Kimi K3'ü 8 GB RAM'li CPU'da çalıştırma iddiası dikkat çekiyor; ancak son 28 gündeki 12 commit'in tek yazardan gelmesi, ilginin bağımsız teknik doğrulama veya geniş katkıcı tabanıyla desteklendiğini göstermiyor.
100,7/gün ort. +191 son gün 0 commit 0 yazar C 01 Ağu 2026 güncellendi 02 Eki 2026
avx2 c99 cpu-inference deep-learning from-scratch inference-engine kimi-k3 linear-attention
9.465
2
mw00/project-maya 2.448
Project Maya - run GLM-5.3-Flash (321B MoE), DeepSeek-V4.1-Flash-1017 (748B MoE) and MiniMax-M3.1-Flash (428B MoE) privately on your own GPU
87,4/gün ort. +55 son gün 897 commit 20 yazar Cuda 06 Eki 2026 güncellendi bugün
cuda glm glm-5 llm llm-inference local-llm mixture-of-experts moe
2.448
3
RyanCodrai/turbovec önem 53gpt6-luna uyum 68gpt6-luna 17.457
turbovec, TurboQuant ile vektörleri sıkıştırıp Rust ve Python üzerinden arama yapmayı sağlayan yerel bir vector index kütüphanesidir.
Neden yükseliyor · 185 günde 17.245 yıldız almış olması ve bellek tasarrufu ile hızlı yerel arama vaadi dikkat çekiyor. Ancak son 28 günde commit ve farklı yazar sayısı sıfır; bu veriler yükselişin güncel olarak sürdüğünü göstermiyor. README'deki performans karşılaştırmaları proje tarafından bildiriliyor, bağımsız doğrulama bilgisi verilen ölçümlerde yok.
78,6/gün ort. +5 son gün 3 commit 1 yazar Rust 26 Mar 2026 güncellendi 27 gün önce
ann avx512 embeddings faiss nearest-neighbor neon python quantization
17.457
4
prabanta-dev/janas 583
37 tok/s from Qwen3-Next-80B (48 GB) on a 32 GB laptop, 24 without drafts vs 2.8 for llama.cpp; reading 62 vs 16 tok/s. Faster than llama.cp
13,3/gün ort. +17 son gün 62 commit 1 yazar C 20 Eyl 2026 güncellendi 4 gün önce
avx2 c cpu-inference gguf inference-engine linux llm local-llm
583
5
whitecircle/halo 840
Halo is an open-source framework built by White Circle for training large language and multimodal models
12,2/gün ort. +1 son gün 5 commit 2 yazar Python 26 Ağu 2026 güncellendi bugün
agent async-rl distributed-training fine-tuning grpo large-language-models llm lora
840
6
Yamz-Labs/kyojin 228
Kyojin: open-source engine that runs the largest open models on one mini PC or one graphics card, about 2× faster than llama.cpp. AMD (ROCm)
7,1/gün ort. +3 son gün 611 commit 5 yazar Python 02 Eki 2026 güncellendi dün
amd exl3 inference llm quantization rocm strix-halo
228
7
Magnvs7/Promethevs 85
PROMETHEVS: run the full DeepSeek-V4.1-Flash-1017 (510 GB) on a gaming PC with 64 GB RAM and a 12 GB GPU. PROMETHEVS engine (compress-then-c
6,5/gün ort. +55 son gün 12 commit 1 yazar 21 Eki 2026 güncellendi 11 gün önce
deepseek llm local-inference mixture-of-experts quantization consumer-hardware deepseek-v4 llm-inference
85
8
leonxger/deepseek-colab 24
Uncensored DeepSeek-V4.1-Flash IQ2_XS on Google Colab
6/gün ort. +10 son gün 1 commit 1 yazar Python 30 Eki 2026 güncellendi 4 gün önce
colab cuda deepseek deepseek-v4 gguf google-colab gpu llama-cpp
24
9
BoumedineBillal/esp32-pie-asr 52
Live English speech recognition on the ESP32-P4: a 65.7M-parameter streaming transducer in 2/4-bit weights, faster than real time, on the ch
5,8/gün ort. +4 son gün 7 commit 1 yazar C++ 25 Eki 2026 güncellendi 2 gün önce
asr embedded esp32-p4 quantization risc-v simd speech-recognition tinyml
52
10
fermionresearch/phonon 327
Phonon: open speech recognition models (Phonon-2, Phonon-1) — CLI, CPU and CUDA images
4,9/gün ort. +3 son gün 7 commit 1 yazar Python 28 Ağu 2026 güncellendi 6 gün önce
asr cuda docker low-bit mlx pytorch quantization speech-recognition
327
11
wanshuiyin/ARIS-in-AI-Offer 632
Bilingual (中文+EN) ML / LLM / diffusion / agent interview cheat sheets for AI 秋招 — generated by ARIS /interview-cheatsheet, rendered by /rend
3,8/gün ort. +3 son gün 4 commit 1 yazar Python 19 May 2026 güncellendi 4 gün önce
ai-interview aris cheatsheet chinese claude-code deep-learning diffusion flow-matching
632
12
JakeATX/llamAmpere 214
llama.cpp fork for significantly improved performance on Ampere generation Nvidia cards (especially RTX 3090 / 3090 Ti): example: 110+ tok/s
3,8/gün ort. +6 son gün 775 commit 2 yazar C++ 07 Eyl 2026 güncellendi bugün
ampere consumer-gpu cuda ggml gguf kv-cache-quantization llama-cpp llm
214
13
zengxiao-he/tessera 558
From teacher to tiles — a from-scratch LLM distillation & serving engine: custom Triton/CUDA kernels, FSDP distillation, paged-KV continuous
3,7/gün ort. +0 son gün 0 commit 0 yazar Python 05 Haz 2026 güncellendi 24 gün önce
cuda flash-attention fsdp inference-engine jax knowledge-distillation kv-cache llm
558
14
huawei-csl/KVarN 513
KVarN is a native vLLM KV-cache quantization backend for your agents: 3-5x more context, throughput above FP16, and FP16-level accuracy. Cal
3,2/gün ort. +0 son gün 0 commit 0 yazar Python 29 May 2026 güncellendi 22 Haz 2026
agentic-ai kv-cache long-context quantization vllm llm llm-inference
513
15
Dreamer-Toby/STEPQuant 101
STEPQuant: When and Where Errors Matter in Delta-Rule Recurrent State Quantization
2,9/gün ort. -1 son gün 2 commit 1 yazar Python 29 Eyl 2026 güncellendi 11 gün önce
large-language-models llm post-training-quantization quantization
101
16
zhangshuibai/molten 23
Run a coding model at 300+ tok/s on your Mac. Native Rust + Metal inference engine for LiquidAI LFM2.5-24B-A2B (MLX 4-bit) on Apple Silicon.
2,6/gün ort. +0 son gün — commit Metal 25 Eki 2026 güncellendi 5 gün önce
apple-silicon gpu-kernels inference-engine inference-optimization lfm2 liquid-ai llama-cpp llm
23
17
AEON-7/Qwen3.6-27B-AEON-Ultimate-Uncensored-DFlash 489
Fully uncensored, capability-enhanced abliteration of Qwen3.6-27B. NVFP4 + z-lab DFlash speculative decoding (n=12) on the unified ghcr.io/a
2,5/gün ort. +1 son gün 0 commit 0 yazar Python 25 Nis 2026 güncellendi 28 gün önce
abliteration blackwell dflash dgx-spark llm nvfp4 quantization qwen
489
18
ronitgupta138/Vectra-Core 64
⚡ In-Memory Vector Search Engine & HNSW ANN Graph in Java 21 Loom. Sub-millisecond KNN retrieval, 8-bit scalar quantization (SQ8), 16,000+ Q
2,1/gün ort. +0 son gün 31 commit 1 yazar Java 04 Eki 2026 güncellendi 4 gün önce
approximate-nearest-neighbors embeddings high-throughput hnsw java-21 machine-learning quantization similarity-search
64
19
giannisanni/pulsar 234
SSD-streaming inference engine for giant MoE models (Rust + CUDA). GLM 5.2 743B at 2 tok/s and Hy3 295B at 7 tok/s on two consumer 16GB GPUs
2,1/gün ort. +3 son gün 0 commit 0 yazar Rust 13 Tem 2026 güncellendi 01 Eyl 2026
cuda gguf glm inference-engine llm local-llm mixture-of-experts moe
234
20
FujitsuResearch/OneCompression 445
Python package for LLM compression
2/gün ort. +0 son gün 17 commit 4 yazar Python 30 Mar 2026 güncellendi dün
llm qep quantization vllm
445
21
0xBakeer/TandemLLM 73
An environment where open models work as agents: a first-level harness with routing, skills, memory and sandboxed tools, on an engine that d
2/gün ort. +3 son gün — commit Python 28 Eyl 2026 güncellendi bugün
dgx-spark llm-inference nvfp4 qwen speculative-decoding ai-agents exl3 inference-engine
73
22
Saganaki22/ComfyUI-Ming-3-tts 28
Native ComfyUI port of Ming-3.0-tts, supporting 0.5B and Tiny 9B speech generation.
2/gün ort. +0 son gün — commit Python 20 Eki 2026 güncellendi 3 gün önce
comfyui ming one-shot quantization text-to-speech tts voice-clone voice-cloning
28
23
quantumaikr/quant.cpp 405
LLM inference with 7x longer context. Pure C, zero dependencies. Lossless KV cache compression + single-header library.
1,8/gün ort. +0 son gün 0 commit 0 yazar C 28 Mar 2026 güncellendi 26 Nis 2026
delta-compression embeddable gguf kv-cache llm llm-inference pure-c quantization
405
24
mattn/tensai 118
A tiny neural-network framework in pure Go with AVX2 SIMD kernels (GOEXPERIMENT=simd)
1,6/gün ort. +0 son gün 11 commit 1 yazar Go 19 Ağu 2026 güncellendi 25 gün önce
autograd avx2 deep-learning gguf golang llm machine-learning neural-networks
118
25
tlack/babytalk 55
Optimized ESP32-S3/P4 fully offline speech to text and text to speech system for Micropython and AtomVM, featuring a finetuned STT model for
1,4/gün ort. +1 son gün — commit C 26 Eyl 2026 güncellendi 20 gün önce
asr atomvm edge-ai elixir erlang esp-idf esp32 esp32-p4
55
26
alleyxd/quantpicker 15
Measures which GGUF quant of a model to run on your GPU: real speed and quality loss (KL divergence), not estimates.
1,4/gün ort. +0 son gün — commit Rust 23 Eki 2026 güncellendi 6 gün önce
benchmark gguf llama-cpp llm local-llm quantization rust
15
27
timtoole02/Camelid 221
Camelid: a Rust-native local inference backend with evidence-gated model compatibility.
1,2/gün ort. +0 son gün 81 commit 3 yazar Rust 01 May 2026 güncellendi 18 gün önce
apple-silicon gguf inference llama llm local-first metal openai-compatible
221
28
alesha-pro/atlas 75
Interactive canvas for taking an LLM apart tensor by tensor: measured INT8/INT4/FP8 error, distributions, spectra and outlier channels for e
1,2/gün ort. +0 son gün — commit TypeScript 30 Ağu 2026 güncellendi 02 Eyl 2026
dataviz interpretability llm quantization visualization
75
29
Lowaltitudeflight/local-llm-tuning-guide 19
A decision library for running local LLMs on multi-GPU machines without NVLink: parallelism, communication, quantization, speculative decodi
1,1/gün ort. +0 son gün — commit Python 16 Eki 2026 güncellendi bugün
benchmarking decision-guide gpu-serving inference-optimization kv-cache llm llm-deployment llmops
19
30
ayutaz/sanoTTS-jp 70
559 K パラメータの日本語 TTS を ESP32-S3 で実時間合成。漢字かな交じり文の形態素解析・アクセント推定まで端末内で走る（M5Stack CoreS3 実機で確認）。推論は依存ゼロの C99、ブラウザ demo あり。arXiv:2608.21378 sanoTT
1/gün ort. +0 son gün — commit Python 27 Ağu 2026 güncellendi dün
embedded esp32 esp32-s3 g2p japanese japanese-tts knowledge-distillation m5stack
70
31
Nhahan/Xevn 13
A 4B decision model built for Korean.
1/gün ort. +0 son gün — commit 21 Eki 2026 güncellendi 11 gün önce
calibration decision-model jev korean llm mlx quantization qwen
13
32
jjang-ai/jangq 232
JANG — GGUF for MLX. YOU MUST USE JANG_Q RUNTIME. Adaptive Mixed-Precision Quantization + Runtime for Apple Silicon
1/gün ort. +0 son gün — commit Python 15 Mar 2026 güncellendi dün
gguf llamacpp llm mlx mlxllm quantization apple-silicon jang-quantization
232
33
BobClawblaw/deepseek-v41-flash-mlx-m5-ultra 14
DeepSeek-V4.1-Flash-1017 on one 256 GB M5 Ultra: 1M context, 70.7-104.9 tok/s decode with DSpark drafts, 1,739-1,867 tok/s cold prompts (8k-
0,9/gün ort. +1 son gün — commit Python 19 Eki 2026 güncellendi 11 gün önce
apple-silicon deepseek deepseek-v4 llm llm-inference long-context m5-ultra metal
14
34
TilelliLab/atome-lm 153
A ternary, zero-heap tiny language model that runs inside a $2 microcontroller — bit-exact Python <-> C99 <-> Cortex-M3 (QEMU) parity. Apach
0,9/gün ort. +1 son gün — commit Python 23 May 2026 güncellendi 08 Tem 2026
qemu tinyml bitnet cortex-m edge-ai embedded esp32 language-model
153
35
Tylogi/TyloQuant 95
Get more intelligence from every bit. Better quantization formats and smarter calibration let larger, stronger models run smoothly on the ha
0,9/gün ort. +1 son gün — commit C++ 24 Tem 2026 güncellendi 4 gün önce
apple-silicon cuda deepseek gemma gguf llm llm-inference llm-quantization
95
36
ExVRAM/exvram-lab 35
Compute-for-memory research lab for running larger LLMs on memory-constrained consumer GPUs. Official website exvram.pro
0,9/gün ort. +0 son gün — commit Python 26 Eyl 2026 güncellendi 10 gün önce
benchmarking blackwell cuda gpu inference llm local-llm low-bit
35
37
FedericoTs/quantprobe 93
Run a 110B on a 2016 PC with 16 GB RAM. Know your tok/s before you download. Placement beats budget: predicts speed + memory fit for any GGU
0,9/gün ort. +1 son gün — commit Python 21 Tem 2026 güncellendi 27 gün önce
gguf inference llama-cpp llm moe quantization local-llm benchmark
93
38
AarambhDevHub/aarambh-studio 112
🦀 Decoder-only LLM built from scratch in pure Rust using Candle — no Python, no PyTorch. Gated DeltaNet + sparse attention, fine-grained MoE
0,8/gün ort. +1 son gün — commit Rust 24 Haz 2026 güncellendi 21 Ağu 2026
ai-agents candle decoder-only deep-learning dora from-scratch gguf grpo
112
39
IST-DASLab/disaggregated-quantization 32
Disaggregated quantization official codebase
0,7/gün ort. +0 son gün — commit Jupyter Notebook 21 Eyl 2026 güncellendi 13 gün önce
llm quantization quantization-aware-training vllm
32
40
IST-DASLab/GSQ 128
Gumbel-Softmax post-training quantization for LLMs (1–3 bit scalar, INT/GGUF-compatible).
0,7/gün ort. +1 son gün — commit Python 12 May 2026 güncellendi 11 Tem 2026
gguf llm post-training-quantization quantization
128
41
chilly23/RP2040-and-ESP32-AI 54
Run a 28.9M-parameter TinyLM on ESP32-S3 with an RP2040 OLED display node for fully local embedded AI inference.
0,7/gün ort. +0 son gün — commit Python 17 Ağu 2026 güncellendi 17 Ağu 2026
arduino c claude-code codex cpp edge-ai embedded-ai embedded-systems
54
42
helasaoudi/llm-inspector 76
The htop for LLM inference see exactly where every GB of VRAM goes and get measured quantization savings.
0,7/gün ort. +0 son gün — commit Python 15 Tem 2026 güncellendi 28 Eyl 2026
gpu-monitoring htop inference llm llm-inference memory-profiler ollama pip
76
43
FareedKhan-dev/glm-5.2-in-c 75
GLM-5.2, a 744 billion parameter mixture of experts model, in a pure C inference engine: quantized to int4, experts streamed from disk, depl
0,7/gün ort. +0 son gün — commit C 15 Tem 2026 güncellendi 15 Tem 2026
c consumer-hardware cpu-inference cuda deep-learning edge-ai glm inference-engine
75
44
Sandermage/sndr_core_engine 133
SNDR Core Engine (Genesis) — vLLM runtime patch-overlay for Qwen3.6 + Gemma4 on consumer NVIDIA (Ampere sm_86, 2× A5000/3090). Qwen3.6-35B-A
0,7/gün ort. +0 son gün — commit Python 17 Nis 2026 güncellendi 01 Eki 2026
cuda llm-inference qwen vllm speculative-decoding tool-calling rtx-3090 awq
133
45
omgitsbase/llmash 37
Ollama-compatible LLM server on a tuned llama.cpp fork. Beats vllm by 40% on single stream decode, and by 20% on multi stream decode, and be
0,6/gün ort. +0 son gün — commit C++ 06 Eyl 2026 güncellendi bugün
abliteration cuda gemma gguf inference-engine llama-cpp llm llm-inference
37
46
Geekgineer/needle-rs 107
600 KB WASM runtime for Cactus Compute's Needle AI tool-calling models, Needle 3, 2 and 1 from one build. Browser, Cloudflare Workers, Node.
0,6/gün ort. +1 son gün — commit Rust 17 May 2026 güncellendi 20 Eyl 2026
ai browser-ai function-calling llm no-std rust tool-calling transformer
107
47
SparkEngineAI/QuantClaw-plugin 115
QuantClaw is a plug-and-play task-type routing quantization plugin for OpenClaw.
0,6/gün ort. +0 son gün — commit TypeScript 22 Nis 2026 güncellendi 27 Nis 2026
agents claude codex harness llm openclaw quantization
115
48
hwdsl2/scribecrate 121
ScribeCrate: open-source, self-hosted Whisper speech-to-text server powered by faster-whisper. OpenAI-compatible transcription and English t
0,6/gün ort. +2 son gün — commit Python 08 Nis 2026 güncellendi dün
inference openai quantization speech-recognition speech-to-text whisper self-hosted ai
121
49
sauravsingla/MemVanta 40
Low-memory C++20 LLM inference runtime for quantized GGUF models on CPU — mmap-backed weights, paged KV cache, Q4/Q8 kernels, and reproducib
0,5/gün ort. +0 son gün — commit C++ 20 Ağu 2026 güncellendi 2 gün önce
benchmarking cpp cpp20 cpu-inference edge-ai gguf inference kv-cache
40
50
Bellman281/esp32-tinyLLM 48
Rust inference engine for a 28.9M-parameter PLE TinyLM on the ESP32-S3 — 9.4 tok/s, faster than the C implementation it was ported from
0,5/gün ort. +0 son gün — commit Rust 02 Ağu 2026 güncellendi 20 Ağu 2026
embedded esp32 esp32-s3 inference-engine llm quantization rust rust-crate
48