Commits

Branch selector

 master

User selector

nvzhihanj

Datepicker

All time

Commit History

Commits on Oct 22, 2024

• Apply min_new_tokens=2 to mixtral-8x7b, address #1777 (#1884)

  nvzhihanj

  authored
  f74d16f

  Copy full SHA for f74d16f

Commits on Jul 18, 2024

• Urgent: Mixtral submission checker error (#1790)

  nvzhihanj

  authored
  2ed9555

  Copy full SHA for 2ed9555

Commits on Jul 16, 2024

• #1777: TEST06 allows [eos_token_id, eos_token_id] case to WAR mixtral (#1778)

  nvzhihanj

  authored
  c60103b

  Copy full SHA for c60103b

Commits on Jul 11, 2024

• Fix loadgen version (#1775)

  nvzhihanj

  authored
  c83565d

  Copy full SHA for c83565d

Commits on May 21, 2024

• Add open_orca preprocessing steps (#1664)

  Show description for f274a55
  nvzhihanj

  authored
  f274a55

  Copy full SHA for f274a55

Commits on Feb 8, 2024

• Fix submission checker and TEST06 for Llama2 (#1616)

  Show description for 0ed5190
  nvzhihanj

  authored
  0ed5190

  Copy full SHA for 0ed5190

• Turn equal issue mode off for TEST06 (#1615)

  Show description for 817dd96
  nvzhihanj

  authored
  817dd96

  Copy full SHA for 817dd96

Commits on Feb 7, 2024

• Enable equal issue mode for LLM benchmarks (#1610)

  Show description for 357ccef
  nvzhihanj

  authored
  357ccef

  Copy full SHA for 357ccef

Commits on Feb 1, 2024

• #1598: fix token and sample logging for Llama2 when accuracy_log_sampling_target is enabled (#1599)

  nvzhihanj

  authored
  cc3daae

  Copy full SHA for cc3daae

Commits on Jan 26, 2024

• #1558 update llama2 reference fp32 accuracy (#1583)

  Show description for 180014a
  

  nvzhihanj

  and

  mrmhodak

  authored
  180014a

  Copy full SHA for 180014a

Commits on Jan 19, 2024

• Fix llama2 token_per_sample typo and number (#1573)

  nvzhihanj

  authored
  8c27e55

  Copy full SHA for 8c27e55

Commits on Aug 18, 2023

• Fix generate_final_report.py (#1485)

  nvzhihanj

  authored
  1bddedf

  Copy full SHA for 1bddedf

Commits on Aug 17, 2023

• Fix submission checker for GPTJ for datacenter+edge (#1483)

  Show description for 448c984
  nvzhihanj

  authored
  448c984

  Copy full SHA for 448c984

Commits on Aug 3, 2023

• GPTJ TEST01 audit config; GPTJ evaluation deduplication; TEST01 rule improvement (#1458)

  Show description for 7fa94e6
  

  nvzhihanj

  and

  mrasquinha-g

  authored
  7fa94e6

  Copy full SHA for 7fa94e6

Commits on Jul 25, 2023

• Update loadgen version generator to 3.1 (#1452)

  nvzhihanj

  authored
  2da0c52

  Copy full SHA for 2da0c52

Commits on Jul 18, 2023

• Add np.int32 dtype support for GPTJ output (#1437)

  nvzhihanj

  authored
  b4b5cf5

  Copy full SHA for b4b5cf5

• Update loadgen version to v3.1 (#1440)

  nvzhihanj

  authored
  6a6a105

  Copy full SHA for 6a6a105

Commits on Jun 13, 2023

• Update ROUGE score according to latest Intel changes

  

  nvzhihanj

  authored and

  pgmpablo157321

  committed
  bcf6b2b

  Copy full SHA for bcf6b2b

• Fix GPTJ-6B reference accuracy and methods in README

  

  nvzhihanj

  authored and

  pgmpablo157321

  committed
  b8565a0

  Copy full SHA for b8565a0

Commits on May 31, 2023

• Add GPU support to GPTJ reference implementation; code format; fix python wheel readme (#1380)

  nvzhihanj

  authored
  38c0dc5

  Copy full SHA for 38c0dc5

Commits on Aug 6, 2020

• Revert rnnt script to use standard label conversion

  nvzhihanj

  committed
  0cf2c57

  Copy full SHA for 0cf2c57

Commits on Aug 4, 2020

• Address review comment

  nvzhihanj

  committed
  08fba19

  Copy full SHA for 08fba19

• Fix RNNT and BERT script

  nvzhihanj

  committed
  bb04fc9

  Copy full SHA for bb04fc9

Commits on Jul 31, 2020

• fix 3dunet syspath

  nvzhihanj

  committed
  bf3c0c5

  Copy full SHA for bf3c0c5