Search results

Question about optimizing away nuisance hyperparameters and the need for a secondary validation set

Status: Open.

#79 In google-research/tuning_playbook;

· kkew3 opened on Dec 11, 2024

Update for Transformers/LLM/SLM

Status: Open.

#77 In google-research/tuning_playbook;

· fzyzcjy opened on Oct 29, 2024

"trial" or "study"

Status: Open.

#76 In google-research/tuning_playbook;

· yeknafar opened on Sep 13, 2024

Missing image "have_we_sampled_enough.png" in FAQ section

Status: Open.

#74 In google-research/tuning_playbook;

· dylanhogg opened on Aug 3, 2024

Request to cover "2-3 weeks upper bound with data-freeze"

Status: Open.

#73 In google-research/tuning_playbook;

· 21kc-caracol opened on Jul 17, 2024

Request to cover "Insensitivity to tuning" in the playbook

Status: Open.

#72 In google-research/tuning_playbook;

· 21kc-caracol opened on Jul 9, 2024

Why avoid gradient accumulation?

Status: Open.

#69 In google-research/tuning_playbook;

· RonanKMcGovern opened on May 30, 2024

Utilize Hyperband/ASHA scheduler? (Willing to PR)

Status: Open.

#27 In google-research/tuning_playbook;

· fzyzcjy opened on Jan 26, 2023

Consider recommending scipy.stats.qmc for quasi-Monte Carlo sequence generation

Status: Open.

#2 In google-research/tuning_playbook;

· shoyer opened on Jan 19, 2023