books/LLMs-from-scratch
1
0
Fork 0
mirror of https://github.com/rasbt/LLMs-from-scratch.git synced 2026-04-10 12:33:42 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix merge conflict

Browse Source
This commit is contained in:
rasbt
2025-07-10 12:52:10 -05:00
parent ede5cac208
commit 420081ae06
3 changed files with 11 additions and 18 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
ch05/10_llm-training-speed/02_opt_multi_gpu_ddp.py
View File

@@ -312,7 +312,7 @@ def generate_and_print_sample(model, device, start_context):
def train_model_simple_with_timing(model, train_loader, val_loader, optimizer, device,
num_epochs, eval_freq, eval_iter, start_context, tokenizer):
num_epochs, eval_freq, eval_iter, start_context):
train_losses, val_losses, track_tokens = [], [], []
total_tokens, global_step, last_tokens = 0, -1, 0
@@ -524,8 +524,6 @@ def main(gpt_config, settings, rank, world_size):
# Train model
##############################
tokenizer = tiktoken.get_encoding("gpt2")
train_losses, val_losses, tokens_seen = train_model_simple_with_timing(
model=model,
train_loader=train_loader,
@@ -536,7 +534,6 @@ def main(gpt_config, settings, rank, world_size):
eval_freq=5,
eval_iter=1,
start_context="Every effort moves you",
tokenizer=tokenizer
)
# NEW: Clean up distributed processes