cronos3k (Gregor Koch)

published an article 4 months ago

Article

vindex-infer — Run LLMs without CUDA, without PyTorch, from flat binary files

cronos3k

•

Apr 16

• 1

published an article 4 months ago

Article

80% of AI LLM Fine-Tuning Compute Is Wasted — Here's How We Proved It

cronos3k

•

Apr 15

• 1

published an article 4 months ago

Article

Querying a Neural Network Like a Database — and Making That Tool Available to Everyone

cronos3k

•

Apr 15

• 1

published an article 5 months ago

Article

9 Hours, 199 Rounds, Zero Errors: Three Local NemoClaw AI Agents Migrated COBOL to Python Using Only Persistent Memory

cronos3k

•

Mar 18

• 1

published an article 6 months ago

Article

Running a 32 Billion Parameter Image Model on “Ancient” V100 GPUs

cronos3k

•

Jan 29

• 2

published an article over 1 year ago

Article

Unleash the Pixel Woodchipper Pro: Aggressively Disrupting Unwanted Image Data (Maybe?)

cronos3k

•

Apr 22, 2025

• 1

published an article over 1 year ago

Article

Reclaiming Your Words 🛡️: Fighting Stealth Watermarks in AI-Generated Text & Why It Matters (A Developer's Perspective)

cronos3k

•

Apr 22, 2025

• 2

Gregor Koch PRO

AI & ML interests

Organizations

vindex-infer — Run LLMs without CUDA, without PyTorch, from flat binary files

80% of AI LLM Fine-Tuning Compute Is Wasted — Here's How We Proved It

Querying a Neural Network Like a Database — and Making That Tool Available to Everyone

9 Hours, 199 Rounds, Zero Errors: Three Local NemoClaw AI Agents Migrated COBOL to Python Using Only Persistent Memory

Running a 32 Billion Parameter Image Model on “Ancient” V100 GPUs

Unleash the Pixel Woodchipper Pro: Aggressively Disrupting Unwanted Image Data (Maybe?)

Reclaiming Your Words 🛡️: Fighting Stealth Watermarks in AI-Generated Text & Why It Matters (A Developer's Perspective)

Gregor Koch PRO

AI & ML interests

Organizations

cronos3k's activity

vindex-infer — Run LLMs without CUDA, without PyTorch, from flat binary files

80% of AI LLM Fine-Tuning Compute Is Wasted — Here's How We Proved It

Querying a Neural Network Like a Database — and Making That Tool Available to Everyone

9 Hours, 199 Rounds, Zero Errors: Three Local NemoClaw AI Agents Migrated COBOL to Python Using Only Persistent Memory

Running a 32 Billion Parameter Image Model on “Ancient” V100 GPUs

Unleash the Pixel Woodchipper Pro: Aggressively Disrupting Unwanted Image Data (Maybe?)

Reclaiming Your Words 🛡️: Fighting Stealth Watermarks in AI-Generated Text & Why It Matters (A Developer's Perspective)