osmarks/nanogpt-experiments
1
0
Fork 0
mirror of https://github.com/osmarks/nanogpt-experiments.git synced 2024-12-18 14:10:28 +00:00
Code Issues Releases Wiki Activity

rename to compile it's shorter

Browse Source
This commit is contained in:
Andrej Karpathy 2023-01-02 01:14:46 +00:00
parent 2febf4463c
commit 35f51974c4
1 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
train.py
View File

@ -60,7 +60,7 @@ lr_decay_iters = 320000 # how many steps to decay the learning rate for
min_lr = 1e-5 # minimum learning rate min_lr = 1e-5 # minimum learning rate
# DDP settings # DDP settings
backend = 'nccl' # 'nccl', 'gloo', etc. backend = 'nccl' # 'nccl', 'gloo', etc.
compile_model = True # use PyTorch 2.0 to compile the model to be faster compile = True # use PyTorch 2.0 to compile the model to be faster
# ----------------------------------------------------------------------------- # -----------------------------------------------------------------------------
# poor man's Configurator. Potentially a bad idea. Example usage: # poor man's Configurator. Potentially a bad idea. Example usage:
# $ python train.py override_file --batch_size=32 # $ python train.py override_file --batch_size=32
@ -165,7 +165,7 @@ if init_from == 'resume':
optimizer.load_state_dict(checkpoint['optimizer']) optimizer.load_state_dict(checkpoint['optimizer'])
# compile the model # compile the model
if compile_model: if compile:
print("compiling the model... (takes a ~minute)") print("compiling the model... (takes a ~minute)")
unoptimized_model = model unoptimized_model = model
model = torch.compile(model) # requires PyTorch 2.0 model = torch.compile(model) # requires PyTorch 2.0