tacit

tacit synced commits to refs/pull/449/merge at tacit/nanochat from mirror 2026-01-31 23:59:50 +00:00

6346ed10ce Merge bfbe965790 into 1ddaad1c1c

1ddaad1c1c nuke midtraining from orbit, it's not as needed now that we have a BOS-aligned dataloader. Also change the README a lot. midtrianing is not yet fully properly erased across the board, but good enough for step 1

348fbb301b fix dataloader for midtrain to never crop data. we can't just throw it away like we do in pretraining

3c3a3d7042 warmdown of 0.5 is slightly better:

Compare 4 commits »

tacit synced commits to refs/pull/475/head at tacit/nanochat from mirror 2026-01-31 23:59:50 +00:00

4e70a2b678 Update nanochat/flash_attention.py

tacit synced commits to master at tacit/nanochat from mirror 2026-01-31 23:59:50 +00:00

1ddaad1c1c nuke midtraining from orbit, it's not as needed now that we have a BOS-aligned dataloader. Also change the README a lot. midtrianing is not yet fully properly erased across the board, but good enough for step 1

348fbb301b fix dataloader for midtrain to never crop data. we can't just throw it away like we do in pretraining

Compare 2 commits »

tacit synced commits to refs/pull/475/merge at tacit/nanochat from mirror 2026-01-31 23:59:50 +00:00

8202283fea Merge 4e70a2b678 into 1ddaad1c1c

4e70a2b678 Update nanochat/flash_attention.py

1ddaad1c1c nuke midtraining from orbit, it's not as needed now that we have a BOS-aligned dataloader. Also change the README a lot. midtrianing is not yet fully properly erased across the board, but good enough for step 1

348fbb301b fix dataloader for midtrain to never crop data. we can't just throw it away like we do in pretraining

Compare 4 commits »

tacit synced commits to refs/pull/393/merge at tacit/nanochat from mirror 2026-01-31 23:59:50 +00:00

8622a3eb89 Merge 89d2741cba into 1ddaad1c1c

1ddaad1c1c nuke midtraining from orbit, it's not as needed now that we have a BOS-aligned dataloader. Also change the README a lot. midtrianing is not yet fully properly erased across the board, but good enough for step 1

348fbb301b fix dataloader for midtrain to never crop data. we can't just throw it away like we do in pretraining

3c3a3d7042 warmdown of 0.5 is slightly better:

4d8dbaf6e0 Fix escape character in README bibtex entry (#454)

Compare 10 commits »

tacit synced commits to refs/pull/59/merge at tacit/nanochat from mirror 2026-01-31 15:49:50 +00:00

0954a569b1 Merge 23393eae83 into 3c3a3d7042

3c3a3d7042 warmdown of 0.5 is slightly better:

4d8dbaf6e0 Fix escape character in README bibtex entry (#454)

3ba42e8135 Fix SDPA KV-cache decode to respect sliding window (#456)

ace6740bdd feat: allow top_k=0 in web api to disable filtering (#458)

Compare 11 commits »

tacit synced commits to refs/pull/455/merge at tacit/nanochat from mirror 2026-01-31 15:49:50 +00:00

542cd614e3 Merge ea86cef249 into 3c3a3d7042

3c3a3d7042 warmdown of 0.5 is slightly better:

Compare 2 commits »

tacit synced commits to refs/pull/400/merge at tacit/nanochat from mirror 2026-01-31 15:49:49 +00:00

7530c4a22b Merge dd9d4f051b into 3c3a3d7042

3c3a3d7042 warmdown of 0.5 is slightly better:

4d8dbaf6e0 Fix escape character in README bibtex entry (#454)

3ba42e8135 Fix SDPA KV-cache decode to respect sliding window (#456)

ace6740bdd feat: allow top_k=0 in web api to disable filtering (#458)

Compare 8 commits »

tacit synced commits to refs/pull/475/merge at tacit/nanochat from mirror 2026-01-31 07:39:50 +00:00

4a1c058039 Merge 2e45b7800a into 3c3a3d7042

3c3a3d7042 warmdown of 0.5 is slightly better:

Compare 2 commits »

tacit synced commits to refs/pull/312/merge at tacit/nanochat from mirror 2026-01-31 07:39:50 +00:00

2d4f5fb49d Merge 26b0941f75 into 3c3a3d7042

3c3a3d7042 warmdown of 0.5 is slightly better:

4d8dbaf6e0 Fix escape character in README bibtex entry (#454)

3ba42e8135 Fix SDPA KV-cache decode to respect sliding window (#456)

ace6740bdd feat: allow top_k=0 in web api to disable filtering (#458)

Compare 8 commits »

tacit synced commits to master at tacit/nanochat from mirror 2026-01-31 07:39:49 +00:00

3c3a3d7042 warmdown of 0.5 is slightly better:

tacit synced commits to refs/pull/455/merge at tacit/nanochat from mirror 2026-01-30 23:29:55 +00:00

111b973136 Merge ea86cef249 into 4d8dbaf6e0

4d8dbaf6e0 Fix escape character in README bibtex entry (#454)

3ba42e8135 Fix SDPA KV-cache decode to respect sliding window (#456)

ace6740bdd feat: allow top_k=0 in web api to disable filtering (#458)

2e17723817 Fix generate() crash when top_k=0 (#467)

Compare 10 commits »

tacit synced commits to refs/pull/449/merge at tacit/nanochat from mirror 2026-01-30 23:29:53 +00:00

021bba1d40 Merge bfbe965790 into 4d8dbaf6e0

4d8dbaf6e0 Fix escape character in README bibtex entry (#454)

3ba42e8135 Fix SDPA KV-cache decode to respect sliding window (#456)

ace6740bdd feat: allow top_k=0 in web api to disable filtering (#458)

2e17723817 Fix generate() crash when top_k=0 (#467)

Compare 7 commits »

tacit synced commits to refs/pull/442/merge at tacit/nanochat from mirror 2026-01-30 23:29:52 +00:00

a9d48afc8f Merge 3bf06802e6 into 4d8dbaf6e0

4d8dbaf6e0 Fix escape character in README bibtex entry (#454)

3ba42e8135 Fix SDPA KV-cache decode to respect sliding window (#456)

ace6740bdd feat: allow top_k=0 in web api to disable filtering (#458)

2e17723817 Fix generate() crash when top_k=0 (#467)

Compare 10 commits »

tacit synced commits to refs/pull/437/merge at tacit/nanochat from mirror 2026-01-30 23:29:52 +00:00

ea27c1f3ff Merge 8cfa0451f4 into 4d8dbaf6e0

4d8dbaf6e0 Fix escape character in README bibtex entry (#454)

3ba42e8135 Fix SDPA KV-cache decode to respect sliding window (#456)

ace6740bdd feat: allow top_k=0 in web api to disable filtering (#458)

2e17723817 Fix generate() crash when top_k=0 (#467)

Compare 7 commits »

tacit synced commits to master at tacit/nanochat from mirror 2026-01-30 23:29:51 +00:00

4d8dbaf6e0 Fix escape character in README bibtex entry (#454)

3ba42e8135 Fix SDPA KV-cache decode to respect sliding window (#456)

ace6740bdd feat: allow top_k=0 in web api to disable filtering (#458)

2e17723817 Fix generate() crash when top_k=0 (#467)

02baa15405 i am feeling in a delete mood today. i need to delete a lot of code. there is too much code and surface area and complexity. ew

Compare 6 commits »

tacit synced commits to refs/pull/370/merge at tacit/nanochat from mirror 2026-01-30 23:29:51 +00:00

456258c164 Merge 9a9b12b1be into 4d8dbaf6e0

4d8dbaf6e0 Fix escape character in README bibtex entry (#454)

3ba42e8135 Fix SDPA KV-cache decode to respect sliding window (#456)

ace6740bdd feat: allow top_k=0 in web api to disable filtering (#458)

2e17723817 Fix generate() crash when top_k=0 (#467)

Compare 7 commits »

tacit synced commits to refs/pull/399/merge at tacit/nanochat from mirror 2026-01-30 23:29:51 +00:00

784151d3a6 Merge 76dac3c7a2 into 4d8dbaf6e0

4d8dbaf6e0 Fix escape character in README bibtex entry (#454)

3ba42e8135 Fix SDPA KV-cache decode to respect sliding window (#456)

ace6740bdd feat: allow top_k=0 in web api to disable filtering (#458)

2e17723817 Fix generate() crash when top_k=0 (#467)

Compare 10 commits »

tacit synced commits to refs/pull/311/merge at tacit/nanochat from mirror 2026-01-30 23:29:51 +00:00

eb0c125160 Merge 81a958350a into 067daa7758

067daa7758 small fix cpu script ty PR #474

6a341f2ecf contiguous views and single HtoD transfer for inputs/targets much cleaner

ebd4d9bbf5 tried muonh, appealing but didn't work out of the box

Compare 4 commits »

tacit synced commits to refs/pull/312/merge at tacit/nanochat from mirror 2026-01-30 23:29:51 +00:00

b7e32846ed Merge 26b0941f75 into 067daa7758

067daa7758 small fix cpu script ty PR #474

6a341f2ecf contiguous views and single HtoD transfer for inputs/targets much cleaner

ebd4d9bbf5 tried muonh, appealing but didn't work out of the box

Compare 4 commits »