Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Adds option to eval against both ground truth and learned similarity function's top 1 result #5

Merged
merged 1 commit into from
Nov 20, 2024
Merged

Adds option to eval against both ground truth and learned similarity function's top 1 result #5

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 3 additions & 3 deletions eval_batch.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -91,7 +91,7 @@

def get_cmd(config_file: str, checkpoint: str, batch_size: int, algorithm: str, limit_eval_to_first_n: int, eval_dtype: str) -> str:
cmd = f"CUDA_VISIBLE_DEVICES=1 python3 eval_from_checkpoint.py --eval_batch_size={batch_size} --limit_eval_to_first_n={limit_eval_to_first_n} "
cmd += f"--include_eval_time --eval_dtype={eval_dtype} "
cmd += f"--include_eval_time --eval_dtype={eval_dtype} --eval_against_brute_force "
cmd += f"--gin_config_file={config_file} --top_k_method={algorithm} --inference_from_ckpt={checkpoint} --master_port=12346"
return cmd

Expand Down Expand Up @@ -127,9 +127,9 @@ def eval(dataset: str, batch_size: int) -> List[str]:


if __name__ == "__main__":
dataset = "amzn-books"
#dataset = "amzn-books"
#dataset = "ml-1m"
#dataset = "ml-20m"
dataset = "ml-20m"
batch_size = 32
result = eval(dataset=dataset, batch_size=batch_size)
print(f"================{dataset}===============")
Expand Down
54 changes: 47 additions & 7 deletions eval_from_checkpoint.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -97,6 +97,7 @@
flags.DEFINE_integer("eval_batch_size", 64, "Batch size for evals.")
flags.DEFINE_boolean("include_eval_time", False, "Please set this to False for strict accuracy checks.")
flags.DEFINE_string("eval_dtype", "", "If non-empty, run eval in this dtype.")
flags.DEFINE_boolean("eval_against_brute_force", False, "If true, eval against brute force.")


FLAGS = flags.FLAGS
Expand Down Expand Up @@ -141,7 +142,9 @@ def train_fn(
limit_eval_to_first_n: int,
include_eval_time: bool,
eval_dtype: str,
eval_against_brute_force: Optional[bool] = None,
dataset_name: str = "ml-20m",
custom_date_str: str = "", # not used
max_sequence_length: int = 200,
local_batch_size: int = 128,
eval_batch_size: int = 128,
Expand Down Expand Up @@ -182,6 +185,10 @@ def train_fn(
if eval_dtype == "bf16":
logging.info("Enabling eval in bf16 to speed up.")

if eval_against_brute_force is None:
eval_against_brute_force = include_eval_time
logging.info(f"Eval against brute force set to {eval_against_brute_force}.")

torch.backends.cuda.matmul.allow_tf32 = enable_tf32
torch.backends.cudnn.allow_tf32 = enable_tf32

Expand Down Expand Up @@ -350,6 +357,20 @@ def _rename_state_dict(
eval_dict_all = None
eval_start_time = time.time()
float_dtype = torch.bfloat16 if main_module_bf16 or eval_bf16 or eval_dtype == "bf16" else None
if eval_against_brute_force:
bf_eval_state = get_eval_state(
model=model.module,
all_item_ids=dataset.all_item_ids,
negatives_sampler=negatives_sampler,
top_k_module_fn=lambda item_embeddings, item_ids: get_top_k_module(
top_k_method="MoLBruteForceTopK",
model=model.module,
item_embeddings=item_embeddings,
item_ids=item_ids,
),
device=device,
float_dtype=float_dtype,
)
eval_state = get_eval_state(
model=model.module,
all_item_ids=dataset.all_item_ids,
Expand All @@ -365,13 +386,29 @@ def _rename_state_dict(
)
for eval_iter, row in enumerate(iter(test_data_loader)):
seq_features, target_ids, target_ratings = movielens_seq_features_from_row(row, device=device, max_output_length=gr_output_length + 1)
eval_dict = eval_metrics_v2_from_tensors(
eval_state, model.module, seq_features, target_ids=target_ids, target_ratings=target_ratings,
user_max_batch_size=eval_user_max_batch_size,
include_full_matrices=False,
include_eval_time=include_eval_time,
dtype=float_dtype,
)

if eval_against_brute_force:
# eval against brute-force based ground truth.
bf_eval_top_k_ids = eval_metrics_v2_from_tensors(
bf_eval_state, model.module, seq_features, target_ids=target_ids, target_ratings=target_ratings,
user_max_batch_size=eval_user_max_batch_size,
include_eval_time=False,
include_eval_top_k_ids=True,
dtype=float_dtype,
)["eval_top_k_ids"][:, 0:1]
eval_dict = eval_metrics_v2_from_tensors(
eval_state, model.module, seq_features, target_ids=bf_eval_top_k_ids, target_ratings=target_ratings,
user_max_batch_size=eval_user_max_batch_size,
include_eval_time=include_eval_time,
dtype=float_dtype,
)
else:
eval_dict = eval_metrics_v2_from_tensors(
eval_state, model.module, seq_features, target_ids=target_ids, target_ratings=target_ratings,
user_max_batch_size=eval_user_max_batch_size,
include_eval_time=include_eval_time,
dtype=float_dtype,
)

if eval_dict_all is None:
eval_dict_all = {}
Expand Down Expand Up @@ -433,6 +470,7 @@ def mp_train_fn(
eval_batch_size: int,
include_eval_time: bool,
eval_dtype: str,
eval_against_brute_force: bool,
) -> None:
if gin_config_file is not None:
# Hack as absl doesn't support flag parsing inside multiprocessing.
Expand All @@ -448,6 +486,7 @@ def mp_train_fn(
eval_user_max_batch_size=eval_batch_size,
include_eval_time=include_eval_time,
eval_dtype=eval_dtype,
eval_against_brute_force=eval_against_brute_force,
)

def main(argv):
Expand All @@ -466,6 +505,7 @@ def main(argv):
FLAGS.eval_batch_size,
FLAGS.include_eval_time,
FLAGS.eval_dtype,
FLAGS.eval_against_brute_force,
),
nprocs=world_size,
join=True)
Expand Down