1.7B LLM running at 32 tok/s on the Qualcomm QCS6490!
We deployed 𝗢𝗽𝘁𝗶 𝟭.𝟳𝗕, our 1.58-bit language model (QAT-trained from Qwen 1.7B), on Hexagon NPU inside the QCS6490. 🔹 Peak memory: ~680 MB (weights + KV cache) 🔹 Response rate(Generation): ~32 tokens/s
▶︎
Black and white abstract screensaver | paintings for Frame TV Smart TV | multiple pictures
▶︎
RL for Agents Workshop - Deep Dive on Training Agents with RL and Open Source
▶︎
How China's Secret Network Got Around America's AI Chip Ban
▶︎
Trump Bulls**ts About Iran Negotiations & Throws ICE Into Airport Chaos Mix | The Daily Show
▶︎
Why AI Tokens are so Expensive - Computerphile
▶︎
Why AI Might Not Replace Your Job After All
▶︎
Why I'm no longer using Stripe
▶︎
This Shouldn’t Be Able to Run 120B Locally
▶︎
Trump Praises "Strong" Talks With Iran to End the War | Vantage with Palki Sharma
▶︎
Scott Ritter: Russland gewinnt den Krieg – und das eindeutig
▶︎
Oil prices plunge after Trump Truth Social post | Ed Conway analysis
▶︎
Prompt Engineering Tutorial – Master ChatGPT and LLM Responses
▶︎
Comedians take swings at Trump’s plea for Iran help | Have I Got News For You
▶︎
But what is a neural network? | Deep learning chapter 1
▶︎
Decoding the Controversial Exit of HDFC’s Chairman | Vantage with Palki Sharma
▶︎
Parteiverbot durch die Hintertür: Gesinnungsprüfung für AfD-Kandidaten – Martin Sichert im Interview
▶︎
Pink Ombre Aura Screen | 3 Hours and 1 Second | No Sound
▶︎
Make Your Computer Your Property Again!
▶︎
GUR Hit Squad. Monaco. Spain.
▶︎