AgentDish directory
optimizer
Accepted listings with this tag.
| Listing | Category | Score | Trend | Checked | |
|---|---|---|---|---|---|
#2183
↑ +6
Training a 3.8B LLM to 0.384 CORE for $998
A detailed technical write-up about training a 3.8B-parameter language model from scratch, including setup, throughput improvements, ablations, benchmarks, and cost/performance results. |
AI Research / LLM Training / Write-up | 78 | ↑ +6 | 46 days ago | Details |
#2578
→ 0
SpectraAdamw
A research-oriented optimizer project claiming to cut AdamW optimizer-state memory for LLM training by 50% through factorized variance and frequency-domain compression. The repo includes a Python POC, benchmark claims, and a stated drop-in usage pattern for PyTorch training. |
Developer Tools / ML Training / Optimizers | 71 | → 0 | 17 days ago | Details |