CO @coreutil
Showing posts older than #87 · Back to latest
Core Ai Tools Photo
> Hermes Agent - Gets smarter every session.
> ZeroClaw - 3.4MB. Boots in 10ms.
> memU - 90% lower token costs. Never forgets.
> NemoClaw - AI sandboxed at infrastructure level.
> FreeClaw - Zero API costs. Ever.
> MaxClaw - One click. Agent is live.
> NanoClaw - Every agent in its own container.
> Nanobot - 4,000 lines. Fully readable.
> NullClaw - 678KB. Runs on a $5 chip.
> MimiClaw - No OS. No Linux. Just a $5 chip.
> OpenFang - Runs on a schedule. Zero prompts.
> PicoClaw - Runs on a $10 board.
> IronClaw - AI never sees your passwords.
> TinyClaw - Multiple agents. One goal.
> Moltworker - No server. Runs on Cloudflare.
Core Ai Tools 13). MaxClaw by MiniMax: AI agent in Telegram/WeChat in 20 sec. No Docker, no API keys. > web search, code, images/video > 200K token memory > scheduled digests 💰 From ~$10/mo, free trial. Ditch OpenClaw headaches. agent.minimax.io #MaxClaw #AIAgent
Core Ai Tools 12. NemoClaw (TypeScript) github.com/NVIDIA/NemoClaw NemoClaw wraps your keys, your files, your credentials. in an NVIDIA OpenShell sandbox. The agent physically cannot reach anything outside its policy boundaries. > Zero plaintext credentials > Network…
Core Ai Tools 11. Hermes Agent (Python) github.com/NousResearch/hermes-agent Every task it completes gets written down as a skill. Next time it faces the same problem it already knows the solution. The longer you run it, the more capable it becomes. It doesn't just remember.…
→ Whisper Large v3 + Flash Attention 2: 150 min of audio in 98 seconds
→ Distil Whisper + Flash Attention 2: 150 min in 78 seconds
→ Standard Whisper without optimization: 31 minutes for the same job
→ That's a 19x speedup. Same model. Same accuracy. Just faster.
→ One command to transcribe any audio file or URL
→ Speaker diarization — knows WHO said WHAT
→ Transcription AND translation to other languages
→ Runs on NVIDIA GPUs and Mac (Apple Silicon)
→ Flash Attention 2 for maximum speed
→ Clean JSON output with timestamps
→ Works with every Whisper model variant
Otter.ai charges $100/year. Rev charges $1.50/minute. Descript charges $24/month. Enterprise transcription contracts cost thousands.
Podcasters, journalists, researchers, lawyers, content creators — anyone still paying for transcription is lighting money on fire.
8.8K GitHub stars. 633 forks. MIT License.
13 abliteration methods, 116 models, 837 tests, and it gets SMARTER every time someone runs it
its called OBLITERATUS
it finds the exact weights that make the model refuse and surgically removes them, full reasoning stays intact, just the refusal disappears
15 analysis modules map the geometry of refusal BEFORE touching a single weight, it can even fingerprint whether a model was aligned with DPO vs RLHF vs CAI just from subspace geometry alone
then it cuts, the model keeps its full brain but loses the artificial compulsion to say no
every time someone runs it with telemetry enabled their anonymous benchmark data feeds a growing community dataset, refusal geometries, method comparisons, hardware profiles at a scale no single lab could build