Search results

[Perf] Training speed issue for HOPE-mid (very slow step time with 8×A100)

Status: Open.

#13 In kmccleary3301/nested_learning;

· Calvert0921 opened on Mar 18, 2026

[Perf] Loss ~6 from 4k–10k steps on pilot_160m_multi.yaml (custom seq_len=1024, dim=712, batch=2); where are eval + baseline comparisons?

Status: Open.

#12 In kmccleary3301/nested_learning;

· nijagalanand1 opened on Feb 16, 2026

[Eval] Benchmark against NanoGPT hackers

Status: Open.

#11 In kmccleary3301/nested_learning;

· TomLucidor opened on Feb 6, 2026

[Faithfulness] Major fault in gradient propagation

Status: Open.

#9 In kmccleary3301/nested_learning;

· weipeilun opened on Dec 27, 2025

Relations between sequence, segment, batch and chunk

Status: Open.

#8 In kmccleary3301/nested_learning;

· embarrassing-noob-questions opened on Dec 25, 2025

creating artifacts/tokenizer/refinedweb_mix/spm_32000_unigram.model

Status: Open.

#6 In kmccleary3301/nested_learning;

· virtustate opened on Nov 29, 2025