tacit/nanochat
1
0
Fork 0
mirror of https://github.com/karpathy/nanochat.git synced 2026-04-06 07:35:32 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity
76 Commits 3 Branches 0 Tags 179 MiB
13b97b6088
Commit Graph

76 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Caleb DeLeeuw
13b97b6088
Merge branch 'karpathy:master' into claude/nanochat-sae-interpretability-011CUT2TocZpFerXthoW9LMf 2025-10-25 19:11:19 -07:00
Andrej Karpathy
c75fe54aa7 readme tweak, link to new discussion and add file structure 2025-10-25 19:39:16 +00:00
Claude
558e949ddd
Add SAE-based interpretability extension for nanochat 
This commit adds a complete Sparse Autoencoder (SAE) based interpretability
extension to nanochat, enabling mechanistic understanding of learned features
at runtime and during training.

## Key Features

- **Multiple SAE architectures**: TopK, ReLU, and Gated SAEs
- **Activation collection**: Non-intrusive PyTorch hooks for collecting activations
- **Training pipeline**: Complete SAE training with dead latent resampling
- **Runtime interpretation**: Real-time feature tracking during inference
- **Feature steering**: Modify model behavior by intervening on features
- **Neuronpedia integration**: Prepare SAEs for upload to Neuronpedia
- **Visualization tools**: Interactive dashboards for exploring features

## Module Structure

```
sae/
├── __init__.py          # Package exports
├── config.py            # SAE configuration dataclass
├── models.py            # TopK, ReLU, Gated SAE implementations
├── hooks.py             # Activation collection via PyTorch hooks
├── trainer.py           # SAE training loop and evaluation
├── runtime.py           # Real-time interpretation wrapper
├── evaluator.py         # SAE quality metrics
├── feature_viz.py       # Feature visualization tools
└── neuronpedia.py       # Neuronpedia API integration

scripts/
├── sae_train.py         # Train SAEs on nanochat activations
├── sae_eval.py          # Evaluate trained SAEs
└── sae_viz.py           # Visualize SAE features

tests/
└── test_sae.py          # Comprehensive tests for SAE implementation
```

## Usage

```bash
# Train SAE on layer 10
python -m scripts.sae_train --checkpoint models/d20/base_final.pt --layer 10

# Evaluate SAE
python -m scripts.sae_eval --sae_path sae_models/layer_10/best_model.pt

# Visualize features
python -m scripts.sae_viz --sae_path sae_models/layer_10/best_model.pt --all_features
```

## Design Principles

- **Modular**: SAE functionality is fully optional and doesn't modify core nanochat
- **Minimal**: ~1,500 lines of clean, hackable code
- **Performant**: <10% inference overhead with SAEs enabled
- **Educational**: Designed to be easy to understand and extend

See SAE_README.md for complete documentation and examples.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>
 2025-10-25 01:22:51 +00:00
Andrej Karpathy
05a051dbe9 fix tokenization bug, there should be no space before first letter. sigh 2025-10-24 15:06:06 +00:00
Andrej Karpathy
8892470f29 add the SpellingBee task so that nanochat can count r in strawberry etc. along the way we had to add a bunch of new functionality, e.g. extend the calculator to support the count function of python. possibly the current TaskMixture uses way too many synthetic examples of SpellingBee because the eval gives us exactly 100% performance on spelling. We can tune this later to reclaim some wall clock time here I think 2025-10-24 14:02:48 +00:00
Andrej Karpathy
81597cd616 move the lr schedule args up in base_train so they are tunable in configurator 2025-10-24 13:27:31 +00:00
Andrej Karpathy
cc3636b01c allow the tokenizer visualize_tokenization to also print the exact token id. you can never be paranoid enough 2025-10-24 13:27:05 +00:00
Andrej Karpathy
5eeb2b6ef9 experiment: looking to 'hire' a nanochat repo czar to help the repo, mentioning in readme 2025-10-22 16:55:54 +00:00
Andrej Karpathy
2dda5c4c8d Merge branch 'ulanch-fix/ios-safari-input-overlap' 2025-10-22 16:26:35 +00:00
Andrej Karpathy
80b203ea59 also bump run1000.sh to new uv sync 2025-10-22 16:25:36 +00:00
Luke Stanley
917c858136 Updates lockfile with CPU package support without overwriting other architectures 2025-10-22 16:25:36 +00:00
Luke Stanley
db1d5b595d Git ignore eval_bundle 2025-10-22 16:25:36 +00:00
Luke Stanley
dd9387b362 Fix GPU-less CPU use on Linux with specific Torch indexes 2025-10-22 16:25:36 +00:00
Luke Stanley
32571664b1 Fix Torch crash caused by pinning on CPU 2025-10-22 16:25:36 +00:00
Andrej Karpathy
51e70f0d3c Merge branch 'lukestanley-fix-cpu-support-with-extras' 2025-10-22 16:11:15 +00:00
Andrej Karpathy
48387cd895 also bump run1000.sh to new uv sync 2025-10-22 16:08:31 +00:00
ulanch
796f84527f fix(ui): prevent iOS Safari toolbar from covering input on initial load 2025-10-21 17:34:40 -07:00
Luke Stanley
7a52f9bfbb Updates lockfile with CPU package support without overwriting other architectures 2025-10-21 23:14:34 +00:00
Luke Stanley
760af62e11 Git ignore eval_bundle 2025-10-21 23:14:34 +00:00
Luke Stanley
901b075605 Fix GPU-less CPU use on Linux with specific Torch indexes 2025-10-21 23:14:16 +00:00
Luke Stanley
defd1246aa Fix Torch crash caused by pinning on CPU 2025-10-21 20:28:10 +00:00
Andrej
2e938530ce
delete spurious torch.empty allocation in adamw 
fix: remove unnecessary tensor allocation in DistAdamW optimizer
 2025-10-21 11:35:17 -07:00
Andrej Karpathy
a088b7a6ec use enable_gqa of pytorch sdpa, allows us to delete some code, didnt realize it's available 2025-10-21 18:07:33 +00:00
Andrej Karpathy
94ee507054 quick fix base eval due to fewshot requirement 2025-10-21 17:56:08 +00:00
Andrej
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. 
add cpu|mps support
 2025-10-21 10:26:04 -07:00
Andrej Karpathy
50bea28ef9 also add readme mention of the cpu mps changes 2025-10-21 17:24:48 +00:00
Andrej Karpathy
5bdc99abfb merge and resolve conflict 2025-10-21 17:19:10 +00:00
Andrej Karpathy
dfcb1c16f1 Merge branch 'master' into cpu-mps-dev 2025-10-21 17:15:53 +00:00
Andrej Karpathy
bb71c64579 fix silly issue in dataloader, this version is much faster and more portable to mps too 2025-10-21 17:12:50 +00:00
karpathy
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... 2025-10-21 10:07:40 -07:00
Andrej
c9ea7a91e2
Add customization instructions to README 
Added a section on customization for nanochat.
 2025-10-21 08:57:10 -07:00
Andrej Karpathy
03cddd9878 actually let's not brick code on git pull. change error to warning 2025-10-21 15:13:25 +00:00
Andrej Karpathy
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 2025-10-21 15:04:58 +00:00
karpathy
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 2025-10-20 10:15:17 -07:00
Andrej
a09ac812ed
toml changes for cpu only install 2025-10-20 07:53:15 -07:00
Sermet Pekin
49cd02f283
fix: remove unnecessary tensor allocation in DistAdamW optimizer 
fix: remove unnecessary tensor allocation in DistAdamW optimizer
 2025-10-20 12:03:26 +03:00
burtenshaw
0abb0fa2e3 add both sides of the source check 2025-10-20 10:44:07 +02:00
burtenshaw
c7ae920a77 add check for linux on cpu 2025-10-20 06:51:52 +02:00
Andrej
0f007889dd
Add MIT License as a file to the project 2025-10-19 17:22:19 -07:00
Andrej
5a879f4947
export NANOCHAT_BASE_DIR so child processes get it too 2025-10-19 17:07:56 -07:00
Andrej Karpathy
c1d2ed1c13 use orig_model in sampling, silly of me to miss this 2025-10-20 00:05:09 +00:00
Andrej Karpathy
2bc521a6de use orig_model in sampling, silly of me to miss this 2025-10-20 00:04:15 +00:00
Andrej Karpathy
9467d83cf2 fix memory leak bug in rust tokenizer ty @mitsuhiko 2025-10-19 23:54:31 +00:00
Tancrède Lepoint
b1443dc98c export NANOCHAT_BASE_DIR so child processes get it too 2025-10-19 14:05:40 -04:00
Andrej
cf2baf9933
fix typo 
Co-authored-by: Tancrède Lepoint <tlepoint@users.noreply.github.com>
 2025-10-17 08:35:41 -07:00
karpathy
e4f9b9c64d revert to previous pyproject.toml 2025-10-17 08:08:16 -07:00
Andrej
e883b1d597
Merge pull request #99 from burtenshaw/cpu-mps-dev-ben 
Add mps and cpu dependency management
 2025-10-17 07:24:38 -07:00
burtenshaw
23b6351c1c add groups and source selection 2025-10-17 12:20:18 +02:00
karpathy
ae02650afe update the midtraining script too 2025-10-16 16:33:17 -07:00
karpathy
df600b6ed5 many small tweaks. base, eval, core work now i think 2025-10-16 15:46:18 -07:00