Commits

Branch selector

 main

User selector

All users

All time

Commit History

Commits on Jan 23, 2024

• Update README.md

  ashtianicode

  authored
  13350b9

  Copy full SHA for 13350b9

Commits on Dec 21, 2023

• More epochs

  edkins

  committed
  ac366ae

  Copy full SHA for ac366ae

• Human analysis script and its results

  edkins

  committed
  e2b72e8

  Copy full SHA for e2b72e8

• fast parallel run experiments, and save runs under dir

  ashtianicode

  committed
  d317505

  Copy full SHA for d317505

• Merge branch 'main' of https://github.com/ashtianicode/gradient_hacking into main

  ashtianicode

  committed
  887f77a

  Copy full SHA for 887f77a

• analysis

  ashtianicode

  committed
  5ee1a6f

  Copy full SHA for 5ee1a6f

• Delete finetuning phase2

  edkins

  committed
  7632783

  Copy full SHA for 7632783

• More runs_giles

  edkins

  committed
  54226af

  Copy full SHA for 54226af

• (Partial) Giles's runs from getting_it_working_1

  edkins

  committed
  6d7f601

  Copy full SHA for 6d7f601

• Make goal/expectation optional

  edkins

  committed
  17c45c4

  Copy full SHA for 17c45c4

• requirements.txt and some fixups

  edkins

  committed
  143e4c2

  Copy full SHA for 143e4c2

• Import random trees into games.json and use prefixes.

  Show description for 90bc322
  edkins

  committed
  90bc322

  Copy full SHA for 90bc322

• Some changes from last time

  edkins

  committed
  a0d0c1f

  Copy full SHA for a0d0c1f

Commits on Dec 3, 2023

• Merge branch 'main' of https://github.com/ashtianicode/gradient_hacking into main

  ashtianicode

  committed
  4edb93a

  Copy full SHA for 4edb93a

• expectations in runs.json

  ashtianicode

  committed
  80d3a1a

  Copy full SHA for 80d3a1a

• Add Giles's finetuned model name to models.py

  Show description for 77cdb10
  edkins

  committed
  77cdb10

  Copy full SHA for 77cdb10

• Some fine tuning fixups

  edkins

  committed
  6682921

  Copy full SHA for 6682921

• Actual finetuning experiment

  edkins

  committed
  16b5f89

  Copy full SHA for 16b5f89

• Merge branch 'main' of https://github.com/ashtianicode/gradient_hacking into main

  ashtianicode

  committed
  1fae7b8

  Copy full SHA for 1fae7b8

• feat:parallel and model config

  ashtianicode

  committed
  0b63dcc

  Copy full SHA for 0b63dcc

• Finetuning bug fixes

  Show description for ca176fa
  edkins

  committed
  ca176fa

  Copy full SHA for ca176fa

• Update hand_craft_experiments.py

  Show description for 1e5890d
  ariel-gil

  committed
  1e5890d

  Copy full SHA for 1e5890d

• pe fixups for finetuning phase0

  edkins

  committed
  80514c5

  Copy full SHA for 80514c5

• Tweak the prompt generation for more likely Chain of Thought and answer encapsulation.

  edkins

  committed
  7ac6ebb

  Copy full SHA for 7ac6ebb

• rename: hand_craft_experiement (vs tree generation)

  ashtianicode

  committed
  595476a

  Copy full SHA for 595476a

• Merge branch 'main' of https://github.com/ashtianicode/gradient_hacking into main

  ashtianicode

  committed
  0ae716e

  Copy full SHA for 0ae716e

• loop through experiments

  ashtianicode

  committed
  57d5899

  Copy full SHA for 57d5899

• rm runs.json

  edkins

  committed
  1903891

  Copy full SHA for 1903891

• Finetuning phase0 - run through language model

  edkins

  committed
  ba1cb0e

  Copy full SHA for ba1cb0e

• Add goal to finetuning phase0

  edkins

  committed
  eb9a3d1

  Copy full SHA for eb9a3d1

• Finetuning experiment

  edkins

  committed
  d8e8bef

  Copy full SHA for d8e8bef

• fix: tree_pretty_print

  ashtianicode

  committed
  849bcc9

  Copy full SHA for 849bcc9

• feat: tree_pretty_print

  ashtianicode

  committed
  3638d14

  Copy full SHA for 3638d14

• feat:tree_pretty_print

  ashtianicode

  committed
  5dabb44

  Copy full SHA for 5dabb44

• Merge branch 'main' of https://github.com/ashtianicode/gradient_hacking into main

  ashtianicode

  committed
  928d0e0

  Copy full SHA for 928d0e0