tacit

tacit synced commits to refs/pull/106/merge at tacit/nanochat from mirror 2025-10-21 22:02:15 +00:00

e82ffb57e4 Merge 18aa43744d into a088b7a6ec

a088b7a6ec use enable_gqa of pytorch sdpa, allows us to delete some code, didnt realize it's available

94ee507054 quick fix base eval due to fewshot requirement

33e8a27f91 Merge karpathy/cpu-mps-dev , adding the ability to run on CPU, on MPS, or on CUDA, with autodetect. Gnarly PR, nonzero chance I broke something.

50bea28ef9 also add readme mention of the cpu mps changes

Compare 26 commits »

tacit synced commits to refs/pull/103/merge at tacit/nanochat from mirror 2025-10-21 22:02:15 +00:00

e8f413b068 Merge 5158f00a8c into bb71c64579

bb71c64579 fix silly issue in dataloader, this version is much faster and more portable to mps too

c9ea7a91e2 Add customization instructions to README

03cddd9878 actually let's not brick code on git pull. change error to warning

fe5aed940b add personality to nanochat. breaks previous code on git pull and requires download of a new file from s3, but there is a helpful error message so hopefully its ok

Compare 5 commits »

tacit synced commits to refs/pull/105/merge at tacit/nanochat from mirror 2025-10-21 22:02:15 +00:00

4615a0f386 Merge 50b236fbcc into c9ea7a91e2

c9ea7a91e2 Add customization instructions to README

03cddd9878 actually let's not brick code on git pull. change error to warning

fe5aed940b add personality to nanochat. breaks previous code on git pull and requires download of a new file from s3, but there is a helpful error message so hopefully its ok

Compare 4 commits »

tacit synced commits to refs/pull/126/merge at tacit/nanochat from mirror 2025-10-21 22:02:15 +00:00

6fba2defd7 Merge 539f42bf89 into 2e938530ce

2e938530ce delete spurious torch.empty allocation in adamw

a088b7a6ec use enable_gqa of pytorch sdpa, allows us to delete some code, didnt realize it's available

94ee507054 quick fix base eval due to fewshot requirement

33e8a27f91 Merge karpathy/cpu-mps-dev , adding the ability to run on CPU, on MPS, or on CUDA, with autodetect. Gnarly PR, nonzero chance I broke something.

Compare 28 commits »

tacit synced commits to refs/pull/122/merge at tacit/nanochat from mirror 2025-10-21 22:02:15 +00:00

6984038e03 Merge 4ed203a3ab into 2e938530ce

2e938530ce delete spurious torch.empty allocation in adamw

a088b7a6ec use enable_gqa of pytorch sdpa, allows us to delete some code, didnt realize it's available

94ee507054 quick fix base eval due to fewshot requirement

33e8a27f91 Merge karpathy/cpu-mps-dev , adding the ability to run on CPU, on MPS, or on CUDA, with autodetect. Gnarly PR, nonzero chance I broke something.

Compare 28 commits »

tacit synced and deleted reference refs/tags/refs/pull/127/merge at tacit/nanochat from mirror 2025-10-21 22:02:14 +00:00

tacit synced commits to master at tacit/nanochat from mirror 2025-10-21 22:02:14 +00:00

2e938530ce delete spurious torch.empty allocation in adamw

a088b7a6ec use enable_gqa of pytorch sdpa, allows us to delete some code, didnt realize it's available

94ee507054 quick fix base eval due to fewshot requirement

33e8a27f91 Merge karpathy/cpu-mps-dev , adding the ability to run on CPU, on MPS, or on CUDA, with autodetect. Gnarly PR, nonzero chance I broke something.

50bea28ef9 also add readme mention of the cpu mps changes

Compare 27 commits »

tacit synced commits to cpu-mps-dev at tacit/nanochat from mirror 2025-10-21 22:02:14 +00:00

50bea28ef9 also add readme mention of the cpu mps changes

5bdc99abfb merge and resolve conflict

dfcb1c16f1 Merge branch 'master' into cpu-mps-dev

bb71c64579 fix silly issue in dataloader, this version is much faster and more portable to mps too

bb786c5560 i shouldnt have committed the lock file, i missed that. revert to the flagship build which is linux. sorry to pollute the repo history...

Compare 16 commits »

tacit synced and deleted reference refs/tags/refs/pull/88/merge at tacit/nanochat from mirror 2025-10-21 22:02:14 +00:00

tacit synced and deleted reference refs/tags/refs/pull/137/merge at tacit/nanochat from mirror 2025-10-21 22:02:14 +00:00

tacit synced and deleted reference refs/tags/refs/pull/123/merge at tacit/nanochat from mirror 2025-10-21 22:02:14 +00:00

tacit synced and deleted reference refs/tags/refs/pull/130/merge at tacit/nanochat from mirror 2025-10-21 13:52:13 +00:00

tacit synced commits to refs/pull/39/merge at tacit/nanochat from mirror 2025-10-21 13:52:13 +00:00

3e4e88000b Merge 03e4bc8e68 into 0f007889dd

0f007889dd Add MIT License as a file to the project

5a879f4947 export NANOCHAT_BASE_DIR so child processes get it too

c1d2ed1c13 use orig_model in sampling, silly of me to miss this

2bc521a6de use orig_model in sampling, silly of me to miss this

Compare 7 commits »

tacit synced and deleted reference refs/tags/refs/pull/108/merge at tacit/nanochat from mirror 2025-10-21 05:42:13 +00:00

tacit synced commits to refs/pull/122/head at tacit/nanochat from mirror 2025-10-20 21:32:15 +00:00

4ed203a3ab remove transformers from toml. add it to gh Workflow. copy common.py from cpu|mps branch to check if gh wf tests are passing

tacit synced commits to refs/pull/88/head at tacit/nanochat from mirror 2025-10-20 21:32:15 +00:00

2e9669e03a upgrading all other files to be able to use cpu/mps as well as cuda. various minor other changes ,e.g. changing max_iterations to num_iterations in sft script for consistency in naming

a09ac812ed toml changes for cpu only install

0abb0fa2e3 add both sides of the source check

c7ae920a77 add check for linux on cpu

Compare 4 commits »

tacit synced commits to refs/pull/6/merge at tacit/nanochat from mirror 2025-10-20 21:32:15 +00:00

6b5f29d56c Merge 6d51049077 into 0f007889dd

0f007889dd Add MIT License as a file to the project

5a879f4947 export NANOCHAT_BASE_DIR so child processes get it too

c1d2ed1c13 use orig_model in sampling, silly of me to miss this

2bc521a6de use orig_model in sampling, silly of me to miss this

Compare 7 commits »

tacit synced commits to refs/pull/122/merge at tacit/nanochat from mirror 2025-10-20 21:32:15 +00:00

470294c87d Merge 4ed203a3ab into 0f007889dd

4ed203a3ab remove transformers from toml. add it to gh Workflow. copy common.py from cpu|mps branch to check if gh wf tests are passing

Compare 2 commits »

tacit synced commits to refs/pull/108/merge at tacit/nanochat from mirror 2025-10-20 21:32:15 +00:00

fdbf7b73b1 Merge 9767451f7e into a09ac812ed

a09ac812ed toml changes for cpu only install

0abb0fa2e3 add both sides of the source check

c7ae920a77 add check for linux on cpu

Compare 4 commits »

tacit synced commits to cpu-mps-dev at tacit/nanochat from mirror 2025-10-20 21:32:15 +00:00

2e9669e03a upgrading all other files to be able to use cpu/mps as well as cuda. various minor other changes ,e.g. changing max_iterations to num_iterations in sft script for consistency in naming

a09ac812ed toml changes for cpu only install

0abb0fa2e3 add both sides of the source check

c7ae920a77 add check for linux on cpu

Compare 4 commits »