wassname/multifit
1
0
Fork 0
mirror of https://github.com/wassname/multifit.git synced 2026-09-09 11:27:26 +08:00
Code Issues Packages Projects Releases Wiki Activity

Add poleval19_full and fix name inference from paths

Browse Source
This commit is contained in:
Piotr Czapla
2019-05-13 03:25:12 +02:00
parent d11deac01a
commit 53eadfc232
1 changed files with 21 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files
ulmfit/__main__.py
+21 -5
View File
@@ -1,5 +1,6 @@
import gc
import os
import re
import pprint
import tarfile
import shutil
@@ -33,6 +34,10 @@ def get_dataset_path(p, dataset_template):
for ds_path in ds.parent.glob(pattern):
yield lang, ds_path
name_re = re.compile("(lstm|qrnn)_(.*)_(lmseed-)?.*\.m")
def folder_name_to_model_name(folder_name):
return name_re.match(folder_name).group(2)
class ULMFiT:
@wraps(LMHyperParams)
def lm(self, dataset_path, **changes):
@@ -81,6 +86,12 @@ class ULMFiT:
print("Adding", f, dest)
tar.add(f, dest)
def poleval19_full(self, base, name=None, num_lm_epochs=6, **kwargs):
clsbase = self.poleval19_init(base, num_lm_epochs=num_lm_epochs, **kwargs)
self.poleval19_seeds(clsbase, seed_name='clsweightseed', **kwargs)
self.poleval19_seeds(clsbase, seed_name='clstrainseed', **kwargs)
def poleval19_init(self, base, name=None, **kwargs):
clstrainseed = clsweightseed = ftseed = lmseed = 0
if "wiki" in base:
@@ -110,17 +121,18 @@ class ULMFiT:
def poleval19_seeds(self, base, seed_name='clsweightseed', model_num=10, **kwargs):
name = folder_name_to_model_name(Path(base).name)
for seed in range(0, model_num, 1):
kwargs[seed_name] = seed
print("Seed: ", seed_name, seed)
self.poleval19_eval(glob=base, **kwargs)
self.poleval19_eval(glob=base, name=name, num_lm_epochs=0, **kwargs)
def poleval19_eval(self, glob, name=None, num_lm_epochs=6, num_cls_epochs=8, bs=160, **kwargs):
if name is None:
name = f"ft{num_lm_epochs}_cl{num_cls_epochs}"
print("Setting name to ", name)
self.eval(glob=glob,
return self.eval(glob=glob,
name=name,
num_lm_epochs=num_lm_epochs,
num_cls_epochs=num_cls_epochs,
@@ -134,6 +146,7 @@ class ULMFiT:
skip_on_error=True, **trn_params):
results = []
model_args = {}
last_model_dir = None
if clsweightseed is not None:
model_args["clsweightseed"] = clsweightseed
if clstrainseed is not None:
@@ -143,7 +156,8 @@ class ULMFiT:
if lmseed is not None:
model_args['lmseed'] = lmseed
data_dir = Path("data").absolute()
if not glob.startswith("data"):
glob=str(glob)
if "data" not in glob and not glob.startswith("/"):
glob = "data/"+glob
for base_model in sorted(data_dir.parent.glob(glob)):
print("Processing", base_model)
@@ -151,8 +165,9 @@ class ULMFiT:
try:
_name = name
if name is None:
_name = base_model.name.replace(".m","").replace("lstm_","").replace("qrnn_","")
_name = folder_name_to_model_name(base_model.name)
params = CLSHyperParams.from_lm(dataset_path, base_model, lang=lang, name=_name, **model_args)
last_model_dir = params.model_dir.relative_to(data_dir.parent)
if (params.model_dir/"cls_best.pth").exists():
print("Evaluating previously trained model")
d = params.validate_cls(label_smoothing_eps=label_smoothing_eps, use_cache=True)
@@ -179,7 +194,8 @@ class ULMFiT:
print(f"Saving result to: {to_csv}")
df.to_csv(to_csv)
if return_df:
return df
return last_model_dir, df
return last_model_dir
def remove_lm_saves(self):
for lm_save in Path("data").glob("**/lm_*.pth"):