• Joined on 2024-05-31
tacit synced commits to refs/pull/400/merge at tacit/nanochat from mirror 2026-01-18 17:33:44 +00:00
63bb5831e2 something i've wanted to do for a while - move all .sh runs to their own directory so they don't pollute root dir
a91743c168 Merge branch 've'
d58fcd9d73 log for jan 17
babde18ce1 small tweaks
Compare 14 commits »
tacit synced commits to master at tacit/nanochat from mirror 2026-01-18 17:33:43 +00:00
63bb5831e2 something i've wanted to do for a while - move all .sh runs to their own directory so they don't pollute root dir
a91743c168 Merge branch 've'
babde18ce1 small tweaks
cf5c9e5b8e resolve a crash for odd depths because FA3 needs head_dim % 8 == 0
413e91aa0f optimal ratio is now around 4
Compare 11 commits »
tacit synced commits to refs/pull/151/merge at tacit/nanochat from mirror 2026-01-18 17:33:43 +00:00
d58fcd9d73 log for jan 17
f9a7e0f111 update the CPU/MPS script to give reasonable results. The model can at least answer that Paris is the capital of France and knows that the sky is blue, for about 40 minutes of training on my macbook. Also fixed a bug that existed due to KVCache bfloat16 dtype assumption
Compare 3 commits »
tacit synced commits to refs/pull/442/merge at tacit/nanochat from mirror 2026-01-18 09:23:53 +00:00
f9a7e0f111 update the CPU/MPS script to give reasonable results. The model can at least answer that Paris is the capital of France and knows that the sky is blue, for about 40 minutes of training on my macbook. Also fixed a bug that existed due to KVCache bfloat16 dtype assumption
Compare 2 commits »
tacit synced commits to master at tacit/nanochat from mirror 2026-01-18 09:23:52 +00:00
d58fcd9d73 log for jan 17
tacit synced commits to refs/pull/409/merge at tacit/nanochat from mirror 2026-01-18 09:23:52 +00:00
96ec51cee2 Merge 7f5992a10e7b5204c775becc826eb0352d061d1c into d58fcd9d73
d58fcd9d73 log for jan 17
Compare 2 commits »
tacit synced commits to refs/pull/437/merge at tacit/nanochat from mirror 2026-01-18 09:23:52 +00:00
d58fcd9d73 log for jan 17
f9a7e0f111 update the CPU/MPS script to give reasonable results. The model can at least answer that Paris is the capital of France and knows that the sky is blue, for about 40 minutes of training on my macbook. Also fixed a bug that existed due to KVCache bfloat16 dtype assumption
Compare 3 commits »
tacit synced commits to refs/pull/393/merge at tacit/nanochat from mirror 2026-01-18 01:13:50 +00:00
f9a7e0f111 update the CPU/MPS script to give reasonable results. The model can at least answer that Paris is the capital of France and knows that the sky is blue, for about 40 minutes of training on my macbook. Also fixed a bug that existed due to KVCache bfloat16 dtype assumption
f5425245f9 more GPU types from PR 147 thanks @Qubitium
2955650327 add detection of device to report more correct mfu for bf16
77a46902e4 Fix WANDB_RUN parameter passing in runcpu.sh (#407)
Compare 13 commits »
tacit synced commits to refs/pull/409/head at tacit/nanochat from mirror 2026-01-18 01:13:50 +00:00
7f5992a10e Add support for ROCm backend in speedrun script
f9a7e0f111 update the CPU/MPS script to give reasonable results. The model can at least answer that Paris is the capital of France and knows that the sky is blue, for about 40 minutes of training on my macbook. Also fixed a bug that existed due to KVCache bfloat16 dtype assumption
f5425245f9 more GPU types from PR 147 thanks @Qubitium
2955650327 add detection of device to report more correct mfu for bf16
77a46902e4 Fix WANDB_RUN parameter passing in runcpu.sh (#407)
Compare 42 commits »
tacit synced commits to refs/pull/409/merge at tacit/nanochat from mirror 2026-01-18 01:13:50 +00:00
ce62400456 Merge 7f5992a10e7b5204c775becc826eb0352d061d1c into f9a7e0f111
7f5992a10e Add support for ROCm backend in speedrun script
f9a7e0f111 update the CPU/MPS script to give reasonable results. The model can at least answer that Paris is the capital of France and knows that the sky is blue, for about 40 minutes of training on my macbook. Also fixed a bug that existed due to KVCache bfloat16 dtype assumption
Compare 3 commits »
tacit synced commits to refs/pull/425/merge at tacit/nanochat from mirror 2026-01-18 01:13:50 +00:00
f9a7e0f111 update the CPU/MPS script to give reasonable results. The model can at least answer that Paris is the capital of France and knows that the sky is blue, for about 40 minutes of training on my macbook. Also fixed a bug that existed due to KVCache bfloat16 dtype assumption
Compare 2 commits »
tacit synced commits to master at tacit/nanochat from mirror 2026-01-18 01:13:49 +00:00
f9a7e0f111 update the CPU/MPS script to give reasonable results. The model can at least answer that Paris is the capital of France and knows that the sky is blue, for about 40 minutes of training on my macbook. Also fixed a bug that existed due to KVCache bfloat16 dtype assumption
tacit synced commits to refs/pull/59/merge at tacit/nanochat from mirror 2026-01-17 17:03:48 +00:00
f5425245f9 more GPU types from PR 147 thanks @Qubitium
2955650327 add detection of device to report more correct mfu for bf16
77a46902e4 Fix WANDB_RUN parameter passing in runcpu.sh (#407)
bbc4413c58 Add high value engine tests for core invariants (33 LoC) (#396)
Compare 20 commits »
tacit synced commits to refs/pull/409/merge at tacit/nanochat from mirror 2026-01-17 17:03:47 +00:00
f5425245f9 more GPU types from PR 147 thanks @Qubitium
2955650327 add detection of device to report more correct mfu for bf16
77a46902e4 Fix WANDB_RUN parameter passing in runcpu.sh (#407)
bbc4413c58 Add high value engine tests for core invariants (33 LoC) (#396)
Compare 12 commits »
tacit synced commits to refs/pull/429/merge at tacit/nanochat from mirror 2026-01-17 17:03:47 +00:00
f5425245f9 more GPU types from PR 147 thanks @Qubitium
2955650327 add detection of device to report more correct mfu for bf16
77a46902e4 Fix WANDB_RUN parameter passing in runcpu.sh (#407)
bbc4413c58 Add high value engine tests for core invariants (33 LoC) (#396)
Compare 9 commits »
tacit synced commits to refs/pull/311/merge at tacit/nanochat from mirror 2026-01-17 17:03:46 +00:00
f5425245f9 more GPU types from PR 147 thanks @Qubitium
2955650327 add detection of device to report more correct mfu for bf16
77a46902e4 Fix WANDB_RUN parameter passing in runcpu.sh (#407)
bbc4413c58 Add high value engine tests for core invariants (33 LoC) (#396)
Compare 9 commits »
tacit synced commits to refs/pull/151/merge at tacit/nanochat from mirror 2026-01-17 17:03:45 +00:00
f5425245f9 more GPU types from PR 147 thanks @Qubitium
2955650327 add detection of device to report more correct mfu for bf16
77a46902e4 Fix WANDB_RUN parameter passing in runcpu.sh (#407)
bbc4413c58 Add high value engine tests for core invariants (33 LoC) (#396)
Compare 12 commits »
tacit synced commits to refs/pull/442/merge at tacit/nanochat from mirror 2026-01-17 08:54:04 +00:00
f5425245f9 more GPU types from PR 147 thanks @Qubitium
2955650327 add detection of device to report more correct mfu for bf16
77a46902e4 Fix WANDB_RUN parameter passing in runcpu.sh (#407)
bbc4413c58 Add high value engine tests for core invariants (33 LoC) (#396)
Compare 12 commits »
tacit synced commits to refs/pull/434/merge at tacit/nanochat from mirror 2026-01-17 08:54:04 +00:00
f5425245f9 more GPU types from PR 147 thanks @Qubitium
2955650327 add detection of device to report more correct mfu for bf16
77a46902e4 Fix WANDB_RUN parameter passing in runcpu.sh (#407)
bbc4413c58 Add high value engine tests for core invariants (33 LoC) (#396)
Compare 12 commits »
tacit synced commits to refs/pull/437/merge at tacit/nanochat from mirror 2026-01-17 08:54:04 +00:00
f5425245f9 more GPU types from PR 147 thanks @Qubitium
2955650327 add detection of device to report more correct mfu for bf16
77a46902e4 Fix WANDB_RUN parameter passing in runcpu.sh (#407)
bbc4413c58 Add high value engine tests for core invariants (33 LoC) (#396)
Compare 9 commits »