LightNews LightNews
qdrddr.bsky.social
@qdrddr.bsky.social
37 followers 67 following 240 posts
https://targeterpro.com/linktr-ee-qdrddr
Posts Replies Media Videos
qdrddr.bsky.social
Save 40% tokens when sending #JSON with TOON to #LLMs that is typically better accuracy vs. raw JSON (depends on the model)
GitHub Repo link in the comment 💬👇
📜 MIT lic
#AI #RAG
October 30, 2025 at 8:33 PM
qdrddr.bsky.social
🍏 #Apple #Embedding Atlas — visualize & explore embeddings interactively via Notebook, CLI, or Streamlit. 📜MIT lic
🔗 Link in first 💬⤵️

Repost 🔁 #AI #LLM #RAG
October 25, 2025 at 1:52 PM
qdrddr.bsky.social
⚙️ #Helm 4 is here after 6 years — native SSA, #WebAssembly #WASM plugins, #OCI installs & better kstatus! Beta already out.
🔗 Link in first 💬⤵️

Repost 🔁 #DevOps #Kubernetes #CloudNative
October 18, 2025 at 2:28 PM
qdrddr.bsky.social
⚙️ #Google #DeepMind is building #RISC-V NPUs for edge AI — low-power, open, and framework-ready #TensorFlow, #JAX, #PyTorch
🔗 Link in first 💬⤵️

Repost 🔁 #AI #EdgeAI #Hardware
October 17, 2025 at 5:06 PM
qdrddr.bsky.social
📊 RaBitQ by #LanceDB cuts storage up to 32× with only ~5% recall loss — boosting speed & efficiency for #AI workloads.
🔗 Link in first 💬⤵️

Repost 🔁 #RAG #VectorDB #VectorStore #SemanticSearch
October 16, 2025 at 8:57 PM
qdrddr.bsky.social
🧩 @surrealdb.com — embeddable in-progress #LPG #GraphDB, #SQL & #JSON docs, just like #SQLite for your #AI app perfect replacement for archived #KuzuDB
Supports #GraphQL, #VectorSearch, Full Text.
Python, Rust, JavaScript, .NET bindings.

🔗 Link in first 💬⤵️
Repost 🔁 #Databases #VectorDB #GraphRAG
October 14, 2025 at 3:43 PM
qdrddr.bsky.social
SWE-Bench-Pro (Commercial Dataset) is a new #leaderboard for #programming tasks. Leaders:
- Claude Opus 4.1
- GPT-5
- Gemini 2.5 Pro

Link in the first comment 💬⤵️
#AI #LLM #SWE-Bench
October 7, 2025 at 1:41 AM
qdrddr.bsky.social
ModernVBERT is a vision model

Suite of compact 250M-parameter vision-language encoders, achieving state-of-the-art performance in this size class, matching the performance of models up to 10x larger.
MIT licensed

#AI #LLM #ModernBERT

huggingface.co/collections/...
October 3, 2025 at 6:24 PM
qdrddr.bsky.social
📝 Auto-generate README files • Language-agnostic • Templates, styles & badges
📜 MIT lic
🔗 Link in first 💬⤵️

Repost 🔁 #AI #LLM #Technology
October 1, 2025 at 9:37 PM
qdrddr.bsky.social
🚀 #GEPA: Automatic #Prompt Optimization by @databricksinc.bsky.social: gpt-oss-120b beats Claude Sonnet 4 (+3%) at ~20x lower cost. Completes with DSPy SIMBA/MIPROv2
📜 MIT lic
🔗 Link in first 💬⤵️

Repost 🔁 #AI #LLM #RAG #PromptEngineering #ContextEngineering
September 27, 2025 at 1:11 PM
qdrddr.bsky.social
📂 #RDF #Graph Gen: synthetic generator from #SHACL shapes. Fills dataset gaps & enables structured benchmarks ⚡
🔗 Link in first 💬⤵️

Repost 🔁 #AI #LLM #KnowledgeGraphs
September 2, 2025 at 10:56 PM
qdrddr.bsky.social
📊 MCP-Universe by #Salesforce: real-world #LLM eval with live #MCP servers. Manages token growth, reduces context bloat & confusion ⚡
🔗 Link in first 💬⤵️

Repost 🔁 #AI #RAG #MachineLearning
September 2, 2025 at 10:49 PM
qdrddr.bsky.social
📊 Google discovered #VectorDB document capacity vs embedding size limits — 500k (512), 1.7m (768), 4m (1024), 107m (3072), 250m (4096 dimensions) ⚡ Cross-encoders/rerankers & multi-vector workarounds
🔗 Link in first 💬⤵️

Repost 🔁 #LLM #AI #RAG
September 2, 2025 at 1:18 AM
qdrddr.bsky.social
📂 #Chonkie: lightweight text #chunking for #RAG — code, semantic, late & #Agentic modes, 🧩 Fine-tuned #BERT or embedding model via API detects semantic shifts ⚡ MIT lic
🔗 Link in first 💬⤵️

Repost 🔁 #LLM #AI
August 29, 2025 at 2:44 PM
qdrddr.bsky.social
🗄️ @tur.so: parallelized #SQLite alternative with #MCP — built-in, no setup needed ⚡ MIT lic
🔗 Link in first 💬⤵️

Repost 🔁 #LLM #AI #RAG #Databases #SQL
August 27, 2025 at 7:35 PM
qdrddr.bsky.social
🧠 Memento: #MCP memory for LLMs — improves frozen #Model responses with case-based retrieval & tool ecosystem ⚡
🔗 Link in first 💬⤵️

Repost 🔁 #LLM #AI #RAG
August 27, 2025 at 2:29 PM
qdrddr.bsky.social
🚀 #Hyperlight: Minimalist Rust #sandbox for safe #WebAssembly — ms-fast, kb-light ⚡ Runs on Windows & Linux 📜 Apache 2.0
🔗 Link in first 💬⤵️

Repost 🔁 #Development #Security
August 26, 2025 at 10:18 PM
qdrddr.bsky.social
🔍 MCP-Agent + semantic tool search — reduces #LLM confusion, saves tokens ⚡ Fixes tool name collision 📜 Apache 2.0 lic
🔗 Link in first 💬⤵️

Repost 🔁 #AI #RAG #MCP #MCP-Client
August 25, 2025 at 5:41 PM
qdrddr.bsky.social
#RealityCheck
August 25, 2025 at 5:15 PM
qdrddr.bsky.social
📖 PageIndex: #RAG doc index — no vectors, no chunking, reasoning-based retrieval 🧠 98.7% FinanceBench 📜 MIT lic
🔗 Link in first 💬⤵️

Repost 🔁 #LLM #AI #KnowledgeGraph
August 25, 2025 at 5:11 PM
qdrddr.bsky.social
🔍 Scira: Minimalistic #AI #searchengine — self-hosted or cloud, with citations ⚡ Powered by Vercel AI SDK. Apache 2.0 Lic
🔗 Link in first 💬⤵️

Repost 🔁 #LLM #RAG
August 25, 2025 at 2:10 PM
qdrddr.bsky.social
📂 MemU: Open-source #memory framework for #AI companions — 92% #Locomo #benchmark, fast & low-cost ⚡
🔗 Link in first 💬⤵️

Repost 🔁 #LLM #RAG
August 24, 2025 at 1:19 PM
qdrddr.bsky.social
📉 #Gartner #2025 #HypeCycle: #GenAI goes down (finally!), #KnowledgeGraphs plateau 📊
🔗Link in first 💬⤵️

Repost 🔁 #LLM #AI #RAG
August 22, 2025 at 7:37 PM
qdrddr.bsky.social
⚡ #LiteLLM #Proxy adds async #S3 #caching — up to 120 RPS, reduces costs 💰
🔗Link in first 💬⤵️

Repost 🔁 #LLM #AI #RAG
August 22, 2025 at 2:55 PM
qdrddr.bsky.social
⚡ #Nvidia #Nemotron Nano 9B — 3-6x faster than #Qwen3 8B, 128k context, multilingual Transformer-Mamba hybrid
Link in the first comment 💬⤵️

#LLM #AI #RAG #OpenSource
August 22, 2025 at 2:03 PM