ngmi's avatar
ngmi
c@mailstr.app
npub14p7f...dzry
hi~
ngmi's avatar
ngmi 5 days ago
converting a react children game to rust (macroquad) using GLM5.3 thinking:max, pi cli and $5 in api allowance. let's see how it goes
ngmi's avatar
ngmi 5 days ago
GLM 5.3 Preview and GLM 5.3 Preview Thinking are now available in preview! GLM 5.3 is built for long-horizon coding, autonomous agent workflows, tool use, and complex engineering tasks. It is not included in the NanoGPT subscription yet, but we hope to add it soon. While in preview, prompts and responses may be logged and used for training. As a preview model, its behavior and availability may change while the release is refined #glm
ngmi's avatar
ngmi 5 days ago
Testing pow in #amethyst
ngmi's avatar
ngmi 6 days ago
Super cool tool to check what LLM actually would run locally. try without installing anything: uvx llmfit #llm
ngmi's avatar
ngmi 1 week ago
Error: Anthropic stream error (overloaded_error): Overloaded ๐Ÿคฌ
ngmi's avatar
ngmi 1 week ago
Anything that's not a website or smartcontract should be done in rust. AI is very good at it, and you get a natural layer of protection on top. View quoted note โ†’
ngmi's avatar
ngmi 1 week ago
DeepSeek V4 Pro 0813 and DeepSeek V4 Pro 0813 Thinking are now available and included in the NanoGPT subscription at 1x included-input usage! This general-availability release is built for coding, tool use, cybersecurity, automation, and long-horizon agent workflows. DeepSeek reports major gains over the preview, with V4 Pro 0813 scoring above Opus 4.8 on Terminal Bench 2.1, Cybergym, DeepSWE, and AutomationBench. For now, these models route directly through DeepSeek, which may log and train on your prompts #deepseek
ngmi's avatar
ngmi 1 week ago
Where on earth can I buy LN for USDC now?
ngmi's avatar
ngmi 1 week ago
Testing LFM2.5 2.6B on mobile, abd it's surprisingly good image #LFM #localllm
โ†‘