Hugging Face's logo
Dipankar Sarkar's picture
🤝 Open to Collab

Dipankar Sarkar PRO

dipankarsarkar
7 1519 296
ai-engineering-at's profile picture ErenAta00's profile picture PulseHugger's profile picture
·
https://www.dipankar.cc

AI & ML interests

Building the AI-native stack. Agents as infrastructure, safety as architecture, performance as plumbing. I publish the receipts: papers, datasets, demos.

Recent Activity

repliedto SeaWolf-AI's post 5 minutes ago
💻 Data-center AI, now on a laptop: POCKET-Darwin-180B We're releasing a 4-bit GGUF build of Darwin-180B-RSI, #1 on seven official Hugging Face leaderboards (self-reported), that runs without a GPU. 📦 360 GB → 111 GB (4-bit GGUF, 4 files) 🖥️ No GPU: one server CPU (16 threads) at 18.4–21.0 tokens/s 💻 RTX 5060 laptop (8 GB VRAM) + 32 GB RAM: 4.17 tokens/s 🧊 128 GB mini PC: whole model in memory, no GPU needed 🎯 MMLU-Pro, 2,000 questions, paired: original 87.65% = 4-bit 87.65% How? · Only ~3B of 180B parameters are active per token (10 of 512 experts) · llama.cpp streams just the needed experts from SSD, so 32 GB RAM is enough · Graft quantization: we took the proven Unsloth UD-Q4_K_XL base build and swapped in only the 300 tensors our RSI training changed (300/300 verified) Under the hood is Model-level Recursive Self-Improvement. The model solves verifiable problems, keeps only its own solutions that check out as correct, and trains on them. No human-written solutions or reasoning traces. Built for teams that can't send data to an external cloud (defense, finance, public sector) to run a top-tier model fully offline. 📝 Article: https://huggingface.co/blog/FINAL-Bench/data-center-ai-now-on-a-laptop-pocket-darwin-180b 🤗 Model: https://huggingface.co/FINAL-Bench/POCKET-Darwin-180B-GGUF 🧬 Original: https://huggingface.co/FINAL-Bench/Darwin-180B-RSI #Darwin #RSI #GGUF #llamacpp #OnDevice #MoE
repliedto SeaWolf-AI's post about 1 hour ago
💻 Data-center AI, now on a laptop: POCKET-Darwin-180B We're releasing a 4-bit GGUF build of Darwin-180B-RSI, #1 on seven official Hugging Face leaderboards (self-reported), that runs without a GPU. 📦 360 GB → 111 GB (4-bit GGUF, 4 files) 🖥️ No GPU: one server CPU (16 threads) at 18.4–21.0 tokens/s 💻 RTX 5060 laptop (8 GB VRAM) + 32 GB RAM: 4.17 tokens/s 🧊 128 GB mini PC: whole model in memory, no GPU needed 🎯 MMLU-Pro, 2,000 questions, paired: original 87.65% = 4-bit 87.65% How? · Only ~3B of 180B parameters are active per token (10 of 512 experts) · llama.cpp streams just the needed experts from SSD, so 32 GB RAM is enough · Graft quantization: we took the proven Unsloth UD-Q4_K_XL base build and swapped in only the 300 tensors our RSI training changed (300/300 verified) Under the hood is Model-level Recursive Self-Improvement. The model solves verifiable problems, keeps only its own solutions that check out as correct, and trains on them. No human-written solutions or reasoning traces. Built for teams that can't send data to an external cloud (defense, finance, public sector) to run a top-tier model fully offline. 📝 Article: https://huggingface.co/blog/FINAL-Bench/data-center-ai-now-on-a-laptop-pocket-darwin-180b 🤗 Model: https://huggingface.co/FINAL-Bench/POCKET-Darwin-180B-GGUF 🧬 Original: https://huggingface.co/FINAL-Bench/Darwin-180B-RSI #Darwin #RSI #GGUF #llamacpp #OnDevice #MoE
reacted to comgen42's post with 🔥 about 2 hours ago
🐻 Kodiak-v0.2-1B is out: an open 1B encoder that makes decisions instead of writing text. Give it a state and typed questions. You get back calibrated answers in one forward pass, and it says "can't tell" when it doesn't know. On tasks it was never trained for (our frozen eval set v0.2): • 0.689 accuracy (3-run mean) vs 0.688 for Qwen3-8B, at ~40× the speed • calibration error 0.085 vs 0.293 • accuracy mode (3 models averaged): 0.706 New in v0.2: grounding checks, picking an assistant's next tool step from API specs, claim verification, and product relevance. Known limits are in the model card. 📦 https://huggingface.co/cortex-agent-llc/kodiak-v0.2-1b 🎯 Accuracy mode: https://huggingface.co/cortex-agent-llc/kodiak-v0.2-1b-accuracy 🕹️ Demo: https://huggingface.co/spaces/comgen42/kodiak-demo 📝 What we built, and what didn't work: https://cortexagent.com/blog/kodiak-v0-2-an-open-1b-decision-model-you-can-download-today
View all activity

Organizations

Skelf Research's profile picture Neul Labs's profile picture Cognisoc's profile picture Incredlabs's profile picture