Update tokenizer.py
Browse files- tokenizer.py +28 -28
tokenizer.py
CHANGED
|
@@ -16,35 +16,35 @@ def _get_repo_file(repo_id_or_path: str, filename: str, revision: Optional[str]
|
|
| 16 |
return local
|
| 17 |
return hf_hub_download(repo_id=repo_id_or_path, filename=filename, revision=revision)
|
| 18 |
|
| 19 |
-
|
| 20 |
-
|
| 21 |
-
|
| 22 |
-
|
| 23 |
-
|
| 24 |
-
|
| 25 |
-
|
| 26 |
-
|
| 27 |
-
|
| 28 |
-
|
| 29 |
-
|
| 30 |
-
|
| 31 |
-
return str(x)
|
| 32 |
-
if isinstance(x, (list, tuple)):
|
| 33 |
-
# prefer first/last string element if present
|
| 34 |
-
for pick in (0, -1):
|
| 35 |
-
try:
|
| 36 |
-
v = x[pick]
|
| 37 |
-
if isinstance(v, str):
|
| 38 |
-
return v
|
| 39 |
-
except Exception:
|
| 40 |
-
pass
|
| 41 |
-
# else join all string elements
|
| 42 |
-
parts = [v for v in x if isinstance(v, str)]
|
| 43 |
-
if parts:
|
| 44 |
-
return " ".join(parts)
|
| 45 |
-
return str(x)
|
| 46 |
-
# final fallback
|
| 47 |
return str(x)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 48 |
|
| 49 |
class SyllabicTokenizerWrapper(PreTrainedTokenizerFast):
|
| 50 |
slow_tokenizer_class = None
|
|
|
|
| 16 |
return local
|
| 17 |
return hf_hub_download(repo_id=repo_id_or_path, filename=filename, revision=revision)
|
| 18 |
|
| 19 |
+
def _coerce_to_str(x):
|
| 20 |
+
# common cases first
|
| 21 |
+
if isinstance(x, str):
|
| 22 |
+
return x
|
| 23 |
+
if isinstance(x, dict):
|
| 24 |
+
for key in ("text", "sentence", "input", "prompt"):
|
| 25 |
+
if key in x and isinstance(x[key], str):
|
| 26 |
+
return x[key]
|
| 27 |
+
# fallback: join any stringy values
|
| 28 |
+
vals = [v for v in x.values() if isinstance(v, str)]
|
| 29 |
+
if vals:
|
| 30 |
+
return " ".join(vals)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 31 |
return str(x)
|
| 32 |
+
if isinstance(x, (list, tuple)):
|
| 33 |
+
# prefer first/last string element if present
|
| 34 |
+
for pick in (0, -1):
|
| 35 |
+
try:
|
| 36 |
+
v = x[pick]
|
| 37 |
+
if isinstance(v, str):
|
| 38 |
+
return v
|
| 39 |
+
except Exception:
|
| 40 |
+
pass
|
| 41 |
+
# else join all string elements
|
| 42 |
+
parts = [v for v in x if isinstance(v, str)]
|
| 43 |
+
if parts:
|
| 44 |
+
return " ".join(parts)
|
| 45 |
+
return str(x)
|
| 46 |
+
# final fallback
|
| 47 |
+
return str(x)
|
| 48 |
|
| 49 |
class SyllabicTokenizerWrapper(PreTrainedTokenizerFast):
|
| 50 |
slow_tokenizer_class = None
|