Spaces:
Runtime error
Runtime error
Upload folder using huggingface_hub
Browse files- run_transformers_training.py +22 -14
run_transformers_training.py
CHANGED
@@ -391,7 +391,8 @@ def load_dataset_with_mapping(dataset_config):
|
|
391 |
return 1
|
392 |
|
393 |
def format_phi_chat(messages, dataset_config):
|
394 |
-
"""Format messages according to phi-4's chat template and dataset config.
|
|
|
395 |
formatted_chat = ""
|
396 |
|
397 |
# Get role templates from config
|
@@ -407,13 +408,13 @@ def format_phi_chat(messages, dataset_config):
|
|
407 |
logger.warning(f"Skipping invalid message format: {message}")
|
408 |
continue
|
409 |
|
410 |
-
content = message.get("content", "")
|
411 |
|
412 |
# Skip empty content
|
413 |
if not content:
|
414 |
continue
|
415 |
|
416 |
-
#
|
417 |
if "[RESEARCH INTRODUCTION]" in content:
|
418 |
# System message
|
419 |
template = roles.get("system", "System: {content}\n\n")
|
@@ -429,7 +430,7 @@ def format_phi_chat(messages, dataset_config):
|
|
429 |
template = roles.get("assistant", "Assistant: {content}\n\n")
|
430 |
formatted_chat += template.format(content=content)
|
431 |
|
432 |
-
return formatted_chat
|
433 |
|
434 |
class SimpleDataCollator:
|
435 |
def __init__(self, tokenizer, dataset_config):
|
@@ -459,17 +460,25 @@ class SimpleDataCollator:
|
|
459 |
self.stats["skipped"] += 1
|
460 |
continue
|
461 |
|
462 |
-
#
|
463 |
-
|
464 |
-
|
465 |
-
|
466 |
-
|
467 |
-
logger.warning(f"Empty formatted chat for paper_id {paper_id}, prompt {prompt_num}")
|
468 |
self.stats["skipped"] += 1
|
469 |
continue
|
470 |
|
471 |
-
#
|
472 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
473 |
|
474 |
# Truncate if needed
|
475 |
if len(input_ids) > self.max_seq_length:
|
@@ -489,8 +498,7 @@ class SimpleDataCollator:
|
|
489 |
|
490 |
# Log first few examples for verification
|
491 |
if self.stats["processed"] <= 3:
|
492 |
-
logger.info(f"Sample {self.stats['processed']}
|
493 |
-
logger.info(f"{formatted_chat[:200]}...")
|
494 |
|
495 |
except Exception as e:
|
496 |
logger.warning(f"Error processing example {paper_id}, prompt {prompt_num}: {str(e)}")
|
|
|
391 |
return 1
|
392 |
|
393 |
def format_phi_chat(messages, dataset_config):
|
394 |
+
"""Format messages according to phi-4's chat template and dataset config.
|
395 |
+
Only formats the conversation structure, preserves the actual content."""
|
396 |
formatted_chat = ""
|
397 |
|
398 |
# Get role templates from config
|
|
|
408 |
logger.warning(f"Skipping invalid message format: {message}")
|
409 |
continue
|
410 |
|
411 |
+
content = message.get("content", "") # Don't strip() - preserve exact content
|
412 |
|
413 |
# Skip empty content
|
414 |
if not content:
|
415 |
continue
|
416 |
|
417 |
+
# Only add role prefixes based on position/content
|
418 |
if "[RESEARCH INTRODUCTION]" in content:
|
419 |
# System message
|
420 |
template = roles.get("system", "System: {content}\n\n")
|
|
|
430 |
template = roles.get("assistant", "Assistant: {content}\n\n")
|
431 |
formatted_chat += template.format(content=content)
|
432 |
|
433 |
+
return formatted_chat
|
434 |
|
435 |
class SimpleDataCollator:
|
436 |
def __init__(self, tokenizer, dataset_config):
|
|
|
460 |
self.stats["skipped"] += 1
|
461 |
continue
|
462 |
|
463 |
+
# Get the pre-tokenized content directly
|
464 |
+
# The content should already be properly tokenized and formatted
|
465 |
+
content = conversations[0].get("content", "")
|
466 |
+
if not content:
|
467 |
+
logger.warning(f"Empty content for paper_id {paper_id}, prompt {prompt_num}")
|
|
|
468 |
self.stats["skipped"] += 1
|
469 |
continue
|
470 |
|
471 |
+
# Convert string of numbers to list of integers if needed
|
472 |
+
if isinstance(content, str):
|
473 |
+
try:
|
474 |
+
# Assuming content is space-separated numbers
|
475 |
+
input_ids = [int(x) for x in content.split()]
|
476 |
+
except ValueError:
|
477 |
+
logger.warning(f"Invalid pre-tokenized content format for paper_id {paper_id}, prompt {prompt_num}")
|
478 |
+
self.stats["skipped"] += 1
|
479 |
+
continue
|
480 |
+
else:
|
481 |
+
input_ids = content
|
482 |
|
483 |
# Truncate if needed
|
484 |
if len(input_ids) > self.max_seq_length:
|
|
|
498 |
|
499 |
# Log first few examples for verification
|
500 |
if self.stats["processed"] <= 3:
|
501 |
+
logger.info(f"Sample {self.stats['processed']} token count: {len(input_ids)}")
|
|
|
502 |
|
503 |
except Exception as e:
|
504 |
logger.warning(f"Error processing example {paper_id}, prompt {prompt_num}: {str(e)}")
|