Initial project setup: J-lens implementation for nanoGPT

- Core J-lens computation (batched per-layer, per-token gradient method)
- Project README with background and experiment plan
- Sync script for meru Docker container deployment
- Upstream nanoGPT model code copied to src/

Architecture: Computes d(log_p(token))/d(residual_stream) averaged
over corpus contexts, replicating Anthropic's Jacobian Lens technique.

1 files changed, 25 insertions, 0 deletions

diff --git a/src/config/finetune_shakespeare.py b/src/config/finetune_shakespeare.py
new file mode 100644
index 0000000..148a4c4
--- /dev/null
+++ b/src/config/finetune_shakespeare.py
@@ -0,0 +1,25 @@
+import time
+
+out_dir = 'out-shakespeare'
+eval_interval = 5
+eval_iters = 40
+wandb_log = False # feel free to turn on
+wandb_project = 'shakespeare'
+wandb_run_name = 'ft-' + str(time.time())
+
+dataset = 'shakespeare'
+init_from = 'gpt2-xl' # this is the largest GPT-2 model
+
+# only save checkpoints if the validation loss improves
+always_save_checkpoint = False
+
+# the number of examples per iter:
+# 1 batch_size * 32 grad_accum * 1024 tokens = 32,768 tokens/iter
+# shakespeare has 301,966 tokens, so 1 epoch ~= 9.2 iters
+batch_size = 1
+gradient_accumulation_steps = 32
+max_iters = 20
+
+# finetune at constant LR
+learning_rate = 3e-5
+decay_lr = False