Unverified Commit 78a54e14 authored by Hailey Schoelkopf's avatar Hailey Schoelkopf Committed by GitHub
Browse files

Log `fewshot_as_multiturn` in results files (#1995)



* log fewshot_as_multiturn in general tracker args

* Update evaluator.py

---------
Co-authored-by: default avatarLintang Sutawika <lintang@eleuther.ai>
parent ead2964e
...@@ -271,6 +271,7 @@ def simple_evaluate( ...@@ -271,6 +271,7 @@ def simple_evaluate(
model_args=model_args, model_args=model_args,
system_instruction=system_instruction, system_instruction=system_instruction,
chat_template=lm.chat_template if apply_chat_template else None, chat_template=lm.chat_template if apply_chat_template else None,
fewshot_as_multiturn=fewshot_as_multiturn,
) )
results = evaluate( results = evaluate(
......
...@@ -48,6 +48,7 @@ class GeneralConfigTracker: ...@@ -48,6 +48,7 @@ class GeneralConfigTracker:
model_name_sanitized: str = None model_name_sanitized: str = None
system_instruction: str = None system_instruction: str = None
system_instruction_sha: str = None system_instruction_sha: str = None
fewshot_as_multiturn: bool = None
chat_template: str = None chat_template: str = None
chat_template_sha: str = None chat_template_sha: str = None
start_time: float = None start_time: float = None
...@@ -80,6 +81,7 @@ class GeneralConfigTracker: ...@@ -80,6 +81,7 @@ class GeneralConfigTracker:
model_args: str, model_args: str,
system_instruction: str, system_instruction: str,
chat_template: str, chat_template: str,
fewshot_as_multiturn: bool,
) -> None: ) -> None:
"""Logs model parameters and job ID.""" """Logs model parameters and job ID."""
self.model_source = model_source self.model_source = model_source
...@@ -91,6 +93,7 @@ class GeneralConfigTracker: ...@@ -91,6 +93,7 @@ class GeneralConfigTracker:
) )
self.chat_template = chat_template self.chat_template = chat_template
self.chat_template_sha = hash_string(chat_template) if chat_template else None self.chat_template_sha = hash_string(chat_template) if chat_template else None
self.fewshot_as_multiturn = fewshot_as_multiturn
def log_end_time(self) -> None: def log_end_time(self) -> None:
"""Logs the end time of the evaluation and calculates the total evaluation time.""" """Logs the end time of the evaluation and calculates the total evaluation time."""
......
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