Skip to content

GitLab

Commit e5ec27d7 authored by Mostofa Patwary's avatar Mostofa Patwary
Browse files

addressed comments

parent 5a6431f5
Show whitespace changes
Inline Side-by-side
Showing with 4 additions and 0 deletions
megatron/text_generation_utils.py
View file @ e5ec27d7
...@@ -190,6 +190,10 @@ def generate_samples_input_from_file(model): ...@@ -190,6 +190,10 @@ def generate_samples_input_from_file(model):
raw_text = None raw_text = None
context_count += 1 context_count += 1
# We added this function to support the tasks evaluation such as squad
# and drop in the https://github.com/EleutherAI/lm-evaluation-harness
# codebase. The lm-evaluation-harness code can now call this function
# similar to their current generate function call used for gpt style models.
def generate_samples_eval(model, context, max_gen_length, eos_token_id): def generate_samples_eval(model, context, max_gen_length, eos_token_id):
# Generate samples for lm evaluation # Generate samples for lm evaluation
# NEED TO THINK ABOUT eos token # NEED TO THINK ABOUT eos token
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment