Menu

IV

Ivan Zdravkov

Author Profile

Claim This Author Profile

Prove ownership by publishing #HashtagPLUS and this profile link on your author page or an article under your byline. A moderator or admin will review the request before it merges into your real HashtagPLUS username.

Sign In To Claim
0 karma0 postsjoined about 5 hours ago

🌐 dev.toSource

A kill-gate is a threshold written down before an experiment runs: if the result crosses it, the idea...

about 2 hours ago

🌐 dev.toSource

Quantising a small model down from FP16 to Q4_K_M on my RTX 5070 Ti more than doubled throughput and...

about 2 hours ago

🌐 dev.toSource

A 7B model is the honest ceiling for LoRA fine-tuning on a 16 GB consumer card, not because anything...

about 2 hours ago

🌐 dev.toSource

The Act never uses the word masking. It does not need to. Three articles, on data governance,...

about 4 hours ago

🌐 dev.toSource

Swap the model and almost nothing changes. Swap the masking layer, the permission model, or the...

about 4 hours ago

🌐 dev.toSource

An internal MCP server does not make a model smarter. It solves a governance problem: without one, an...

about 4 hours ago

🌐 dev.toSource

An internal MCP server does not make a model smarter. It solves a governance problem: without one, an...

about 4 hours ago

🌐 dev.toSource

Asked to score the same response five times, a local LLM judge gave the identical answer 36% of the...

about 5 hours ago

🌐 dev.toSource

A small network lost 12.3 percentage points of accuracy on a task it had already mastered, simply by...

about 5 hours ago

🌐 dev.toSource

A ResNet-34 trained on CIFAR-100 is measurably overconfident: its top prediction bin claims 99%...

about 5 hours ago