known-good DA3 training sample (task56) for eval verification

add eval I/O README

da3 single-task finetune of laying_tile_floors (task98), step 44999

da3 single-task finetune of installing_a_modem (task77), step 19999

add eval I/O contract README

add eval I/O contract README

da3 single-task finetune of putting_away_toys (task54), step 34999

da3 single-task finetune of make_rose_centerpieces (task56), step 19999

da3 single-task finetune of make_rose_centerpieces (task56), step 19999

da3 single-task finetune of make_rose_centerpieces (task56), step 19999

100-task meta -> 4-task finetune (54/56/77/98), terminal checkpoint 24999

add fast_tokenizer to task77 assets

100t meta -> single-task ft: installing_a_modem (task77, no-DA3); eval release

add fast_tokenizer to task54 assets

100t meta -> single-task ft: putting_away_toys (task54, no-DA3); eval release

add fast_tokenizer to task56 assets (self-contained eval)

100t meta -> single-task ft: make_rose_centerpieces (task56, no-DA3); eval release

50-task meta -> 4-task finetune (params + qvelfix assets + README; eval release)

Move meta100-1epoch checkpoints to private b1k-meta-checkpoints

Add files using upload-large-folder tool

Add files using upload-large-folder tool

no-DA3 task5 baseline (control arm for T5-DA3-#2), terminal checkpoint 31999

T5-DA3-#2: eval README, source bundles, and the DA3 eval-input builder

DA3-#2 (bundle A + S1) task5 terminal checkpoint, step 32999

Add files using upload-large-folder tool

Add files using upload-large-folder tool

Add files using upload-large-folder tool

Add files using upload-large-folder tool

Add files using upload-large-folder tool

Delete _speedtest/hftest.bin with huggingface_hub

Upload _speedtest/hftest.bin with huggingface_hub

add 100-task 2026 norm_stats (robot-frame qvel, per-row action convention)

task5_setting_mousetraps_da3: params (step 19999)

task5_setting_mousetraps_da3: README, code, assets, bank_mean

add README for task54_putting_away_toys

add README for task56_make_rose_centerpieces

add task54_putting_away_toys step 19999 (params + assets incl. the norm_stats it trained with)

add task56_make_rose_centerpieces step 19999 (params + assets incl. the norm_stats it trained with)

legacy: add index of archived experimental checkpoints

legacy: remove pi05_da3_freezebase_desk_step8000 (now at legacy/pi05_da3_freezebase_desk_step8000)

legacy: remove vggt_newbank_boxing_gloves_step8000 (now at legacy/vggt_newbank_boxing_gloves_step8000)

legacy: remove vggt_newbank_boxing_gloves_step19999 (now at legacy/vggt_newbank_boxing_gloves_step19999)

legacy: remove task12_da3_large_gtdepth_v2_pointmap_step29999_ext30k (now at legacy/task12_da3_large_gtdepth_v2_pointmap_step29999_ext30k)

legacy: remove task12_da3_large_gtdepth_v2_pointmap_step19999 (now at legacy/task12_da3_large_gtdepth_v2_pointmap_step19999)

legacy: remove task12_da3_large_gtdepth_v2_1p5x_slowbase_step19999 (now at legacy/task12_da3_large_gtdepth_v2_1p5x_slowbase_step19999)

legacy: remove task12_da3_large_gtdepth_newbank_step19999 (now at legacy/task12_da3_large_gtdepth_newbank_step19999)

legacy: remove task12_da3_giant_gtdepth_v2_1p5x_step19999 (now at legacy/task12_da3_giant_gtdepth_v2_1p5x_step19999)

legacy: remove kvsplit_strongbase_step14999 (now at legacy/kvsplit_strongbase_step14999)

legacy: remove kvsplit_spatretrain_step19999 (now at legacy/kvsplit_spatretrain_step19999)

legacy: remove kvsplit_spatretrain_step18000 (now at legacy/kvsplit_spatretrain_step18000)