fix: read prompts from config
Browse filesSigned-off-by: Mohammad Kalim Akram <kalim.akram@jina.ai>
- configuration_xlm_roberta.py +2 -0
- modeling_lora.py +11 -10
configuration_xlm_roberta.py
CHANGED
@@ -23,6 +23,7 @@ class XLMRobertaFlashConfig(PretrainedConfig):
|
|
23 |
use_cache=True,
|
24 |
classifier_dropout=None,
|
25 |
lora_adaptations=None,
|
|
|
26 |
lora_rank=4,
|
27 |
lora_dropout_p=0.0,
|
28 |
lora_alpha=1,
|
@@ -55,6 +56,7 @@ class XLMRobertaFlashConfig(PretrainedConfig):
|
|
55 |
self.classifier_dropout = classifier_dropout
|
56 |
self.load_trained_adapters = load_trained_adapters
|
57 |
self.lora_adaptations = lora_adaptations
|
|
|
58 |
self.lora_rank = lora_rank
|
59 |
self.lora_dropout_p = lora_dropout_p
|
60 |
self.lora_alpha = lora_alpha
|
|
|
23 |
use_cache=True,
|
24 |
classifier_dropout=None,
|
25 |
lora_adaptations=None,
|
26 |
+
lora_prompts=None,
|
27 |
lora_rank=4,
|
28 |
lora_dropout_p=0.0,
|
29 |
lora_alpha=1,
|
|
|
56 |
self.classifier_dropout = classifier_dropout
|
57 |
self.load_trained_adapters = load_trained_adapters
|
58 |
self.lora_adaptations = lora_adaptations
|
59 |
+
self.lora_prompts = lora_prompts
|
60 |
self.lora_rank = lora_rank
|
61 |
self.lora_dropout_p = lora_dropout_p
|
62 |
self.lora_alpha = lora_alpha
|
modeling_lora.py
CHANGED
@@ -228,6 +228,14 @@ class XLMRobertaLoRA(XLMRobertaPreTrainedModel):
|
|
228 |
raise ValueError(
|
229 |
f'`lora_adaptations` must be a list and contain at least one element'
|
230 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
231 |
self._adaptation_map = {
|
232 |
name: idx for idx, name in enumerate(self._lora_adaptations)
|
233 |
}
|
@@ -244,13 +252,6 @@ class XLMRobertaLoRA(XLMRobertaPreTrainedModel):
|
|
244 |
self._task_idx = None
|
245 |
# By default, disable LoRA until it's specified which adapter/task to use
|
246 |
self.current_task = None
|
247 |
-
self.prompts = {
|
248 |
-
'query': 'Represent the query for retrieving supporting documents: ',
|
249 |
-
'document': 'Represent the document for retrieval: ',
|
250 |
-
'sts': 'Represent the text for Semantic Textual Similarity: ',
|
251 |
-
'clustering': 'Cluster the text: ',
|
252 |
-
'classification': 'Classify the text: ',
|
253 |
-
}
|
254 |
|
255 |
@property
|
256 |
def main_params_trainable(self):
|
@@ -342,7 +343,7 @@ class XLMRobertaLoRA(XLMRobertaPreTrainedModel):
|
|
342 |
else:
|
343 |
input_ids = kwargs["input_ids"]
|
344 |
input_text = self.roberta.tokenizer.decode(input_ids[0], skip_special_tokens=True)
|
345 |
-
for task_name, prompt in self.
|
346 |
if input_text.startswith(prompt):
|
347 |
self.current_task = task_name
|
348 |
break
|
@@ -385,7 +386,7 @@ class XLMRobertaLoRA(XLMRobertaPreTrainedModel):
|
|
385 |
self.current_task = task_type
|
386 |
else: # infer the task from the input text
|
387 |
input_text = args[0][0] if isinstance(args[0], list) else args[0] # take only the first sentence
|
388 |
-
for task_name, prompt in self.
|
389 |
if input_text.startswith(prompt):
|
390 |
self.current_task = task_name
|
391 |
break
|
@@ -397,4 +398,4 @@ class XLMRobertaLoRA(XLMRobertaPreTrainedModel):
|
|
397 |
)
|
398 |
self.current_task = None # No task-specific adapter is found, just use the general-purpose weights
|
399 |
|
400 |
-
return self.roberta.encode(*args, **kwargs)
|
|
|
228 |
raise ValueError(
|
229 |
f'`lora_adaptations` must be a list and contain at least one element'
|
230 |
)
|
231 |
+
self._lora_prompts = config.lora_prompts
|
232 |
+
if (
|
233 |
+
not isinstance(self._lora_prompts, dict)
|
234 |
+
or len(self._lora_prompts) != len(self._lora_adaptations)
|
235 |
+
):
|
236 |
+
raise ValueError(
|
237 |
+
f'`lora_prompts` must be a dict and contain the same number of elements as `lora_adaptations`'
|
238 |
+
)
|
239 |
self._adaptation_map = {
|
240 |
name: idx for idx, name in enumerate(self._lora_adaptations)
|
241 |
}
|
|
|
252 |
self._task_idx = None
|
253 |
# By default, disable LoRA until it's specified which adapter/task to use
|
254 |
self.current_task = None
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
255 |
|
256 |
@property
|
257 |
def main_params_trainable(self):
|
|
|
343 |
else:
|
344 |
input_ids = kwargs["input_ids"]
|
345 |
input_text = self.roberta.tokenizer.decode(input_ids[0], skip_special_tokens=True)
|
346 |
+
for task_name, prompt in self._lora_prompts.items():
|
347 |
if input_text.startswith(prompt):
|
348 |
self.current_task = task_name
|
349 |
break
|
|
|
386 |
self.current_task = task_type
|
387 |
else: # infer the task from the input text
|
388 |
input_text = args[0][0] if isinstance(args[0], list) else args[0] # take only the first sentence
|
389 |
+
for task_name, prompt in self._lora_prompts.items():
|
390 |
if input_text.startswith(prompt):
|
391 |
self.current_task = task_name
|
392 |
break
|
|
|
398 |
)
|
399 |
self.current_task = None # No task-specific adapter is found, just use the general-purpose weights
|
400 |
|
401 |
+
return self.roberta.encode(*args, **kwargs)
|