Sahil Kadadekar

Crusadersk

https://chimeraforge.vercel.app/reports

AI & ML interests

LLM inference, quantization, refusal robustness, model evaluation, GPU serving, batching effects, post-training, MLOps, AI safety. Reports: https://chimeraforge.vercel.app/reports

Recent Activity

updated a model about 9 hours ago

Crusadersk/qwen2.5-1.5b-medmcqa-drgrpo-lora

published a model about 9 hours ago

Crusadersk/qwen2.5-1.5b-medmcqa-drgrpo-lora

updated a model 9 days ago

Crusadersk/qwen2.5-1.5b-awq-4bit

View all activity

Organizations

New activity in build-small-hackathon/quantsafe-certifier 2 months ago

Correct Codex attribution + rearrange agent trace (rebased on current main; supersedes #27)

#28 opened 2 months ago by

Crusadersk

Correct Codex attribution + rearrange agent trace; sponsors at-a-glance block

#27 opened 2 months ago by

Crusadersk

Top-load sponsors/prizes/badges at-a-glance block

#26 opened 2 months ago by

Crusadersk

Deploy audited QuantSafe Certifier

#25 opened 2 months ago by

Crusadersk

Sync production release trace

#24 opened 2 months ago by

Crusadersk

Deploy audited QuantSafe Certifier

#23 opened 2 months ago by

Crusadersk

Sync final production trace

#22 opened 2 months ago by

Crusadersk

Deploy audited QuantSafe Certifier

#21 opened 2 months ago by

Crusadersk

Publish agent build trace (Claude + Codex passes)

#20 opened 2 months ago by

Crusadersk

Deploy audited QuantSafe Certifier

#19 opened 2 months ago by

Crusadersk

Add NVIDIA NemoGuard Safety-Guard-8B-v3 as 3rd safety judge: Fleiss kappa 0.7929 RELIABLE (up from Cohen 0.7484); sponsor:nvidia; per-model 32B framing

#18 opened 2 months ago by

Crusadersk

Raise certificate JSON contrast

#17 opened 2 months ago by

Crusadersk

Fix dark-preference contrast and component palette

#16 opened 2 months ago by

Crusadersk

Adversarially harden QuantSafe release

#15 opened 2 months ago by

Crusadersk

Replace stale demo with current production cut

#14 opened 2 months ago by

Crusadersk

Reduce ZeroGPU quota per exploratory run

#13 opened 2 months ago by

Crusadersk

Keep pinned Gradio runtime logs clean

#12 opened 2 months ago by

Crusadersk

Batch live probes within ZeroGPU quota

#11 opened 2 months ago by

Crusadersk

Add Llama Champion evidence and fit ZeroGPU quota

#10 opened 2 months ago by

Crusadersk

Run exploratory probe in one ZeroGPU allocation

#9 opened 2 months ago by

Crusadersk

Sahil Kadadekar

AI & ML interests

Recent Activity

Organizations

Crusadersk's activity

Correct Codex attribution + rearrange agent trace (rebased on current main; supersedes #27)

Correct Codex attribution + rearrange agent trace; sponsors at-a-glance block

Top-load sponsors/prizes/badges at-a-glance block

Deploy audited QuantSafe Certifier

Sync production release trace

Deploy audited QuantSafe Certifier

Sync final production trace

Deploy audited QuantSafe Certifier

Publish agent build trace (Claude + Codex passes)

Deploy audited QuantSafe Certifier

Add NVIDIA NemoGuard Safety-Guard-8B-v3 as 3rd safety judge: Fleiss kappa 0.7929 RELIABLE (up from Cohen 0.7484); sponsor:nvidia; per-model 32B framing

Raise certificate JSON contrast

Fix dark-preference contrast and component palette

Adversarially harden QuantSafe release

Replace stale demo with current production cut

Reduce ZeroGPU quota per exploratory run

Keep pinned Gradio runtime logs clean

Batch live probes within ZeroGPU quota

Add Llama Champion evidence and fit ZeroGPU quota

Run exploratory probe in one ZeroGPU allocation