File size: 13,481 Bytes
eae7bce | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 | import json
import re
import warnings
from abc import ABC, abstractmethod
from typing import Dict, List, Tuple
import torch
from transformers import PreTrainedTokenizer
from .template import ChatTemplate
__all__ = ["GeneralParser", "HarmonyParser"]
class Parser(ABC):
def __init__(self, tokenizer: PreTrainedTokenizer, chat_template: ChatTemplate):
self.tokenizer = tokenizer
self.chat_template = chat_template
@abstractmethod
def parse(
self, conversation: "Conversation", max_length: int
) -> Tuple[torch.Tensor, torch.Tensor]:
"""
Parse the conversation into a list of tensors.
Args:
conversation: The conversation to parse.
Returns:
A list of tensors: [input_ids, loss_mask]
"""
_harmony_encoding = None
class GeneralParser(Parser):
def __init__(self, tokenizer: PreTrainedTokenizer, chat_template: ChatTemplate):
super().__init__(tokenizer, chat_template)
self.system_prompt = chat_template.system_prompt
self.user_message_separator = f"{chat_template.end_of_turn_token}"
self.assistant_message_separator = f"{chat_template.assistant_header}"
self.set_assistant_pattern(chat_template)
def apply_chat_template(self, messages, **kwargs) -> str:
conversation = self.tokenizer.apply_chat_template(
messages, tokenize=False, add_generation_prompt=False, **kwargs
)
return conversation
def set_assistant_pattern(self, chat_template: ChatTemplate):
if chat_template.assistant_pattern_type == "longcat":
self.assistant_pattern = (
re.escape(self.assistant_message_separator)
+ r"([\s\S]*?(?:"
+ re.escape("[Round ")
+ r"\d+"
+ re.escape("] USER:")
+ "|$))"
)
else:
self.assistant_pattern = (
re.escape(self.assistant_message_separator)
+ r"([\s\S]*?(?:"
+ re.escape(self.chat_template.end_of_turn_token)
+ "|$))"
)
def parse(
self,
conversation: "Conversation",
max_length: int,
preformatted: bool = False,
train_only_last_turn: bool = False,
**kwargs,
) -> Dict[str, List[torch.Tensor]]:
if not preformatted:
messages = []
if conversation[0]["role"] == "system":
warnings.warn(
f"The first message is from system, we will use the system prompt from the data and ignore the system prompt from the template"
)
messages.append(
{"role": "system", "content": conversation[0]["content"]}
)
conversation = conversation[1:]
else:
if self.system_prompt:
messages.append({"role": "system", "content": self.system_prompt})
for j, sentence in enumerate(conversation):
role = sentence["role"]
if j == 0:
if role != "user":
warnings.warn(
f"Conversation must start with a 'user' role, but found '{role}'. Conversation truncated."
)
break
else:
prev_role = conversation[j - 1]["role"]
if role == "tool" and prev_role not in ["assistant", "tool"]:
warnings.warn(
f"A 'tool' message must follow an 'assistant' or 'tool' message, but was preceded by '{prev_role}'. Conversation truncated."
)
break
if role == "assistant" and prev_role not in ["user", "tool"]:
warnings.warn(
f"An 'assistant' message must follow a 'user' or 'tool' message, but was preceded by '{prev_role}'. Conversation truncated."
)
break
tool_calls = sentence.get("tool_calls")
if isinstance(tool_calls, str):
try:
sentence["tool_calls"] = json.loads(tool_calls)
except json.JSONDecodeError:
warnings.warn(f"Failed to parse tool_calls JSON: {tool_calls}")
break
messages.append(sentence)
try:
conversation = self.apply_chat_template(messages, **kwargs)
except (ValueError, TypeError):
# Fallback rendering for tokenizers without built-in chat_template
warnings.warn(
"Tokenizer does not have a chat_template, using fallback rendering."
)
parts = []
bos_token = getattr(self.tokenizer, "bos_token", None)
user_header = self.chat_template.user_header or ""
assistant_header = self.chat_template.assistant_header or ""
end_of_turn = self.chat_template.end_of_turn_token or ""
# Add BOS token at the start
if bos_token:
parts.append(bos_token)
for msg in messages:
if msg["role"] == "system":
parts.append(msg["content"])
elif msg["role"] == "user":
parts.append(f"{user_header}{msg['content']}")
elif msg["role"] == "assistant":
parts.append(f"{assistant_header}{msg['content']}{end_of_turn}")
conversation = "".join(parts)
if not self.tokenizer.pad_token_id:
self.tokenizer.pad_token_id = self.tokenizer.unk_token_id
# get input_ids
encoding = self.tokenizer(
conversation,
max_length=max_length,
truncation=True,
return_tensors="pt",
add_special_tokens=False,
)
input_ids = encoding.input_ids[0]
loss_mask = torch.zeros(len(input_ids), dtype=torch.long)
matches = list(re.finditer(self.assistant_pattern, conversation, re.DOTALL))
if train_only_last_turn and matches:
matches = [matches[-1]] # Only keep the last match
for match in matches:
content_start_char = match.start(1)
content_end_char = match.end(1)
# --- Core Alternative Operation: Calculate Token Index Based on Prefix String Length ---
# Encode the text "assistant start", the length of which is the position of the starting token.
prefix_ids = self.tokenizer.encode(
conversation[:content_start_char],
add_special_tokens=False,
truncation=True,
max_length=max_length,
)
# Encodes the text "assistant end", the length of which is the position of the end token.
full_ids = self.tokenizer.encode(
conversation[:content_end_char],
add_special_tokens=False,
truncation=True,
max_length=max_length,
)
start_token_idx = len(prefix_ids)
end_token_idx = len(full_ids)
# Handling out-of-bounds errors caused by truncation
actual_start = min(start_token_idx, len(input_ids))
actual_end = min(end_token_idx, len(input_ids))
if actual_start < actual_end:
loss_mask[actual_start:actual_end] = 1
return input_ids, loss_mask
class HarmonyParser(Parser):
def __init__(self, tokenizer: PreTrainedTokenizer, chat_template: ChatTemplate):
super().__init__(tokenizer, chat_template)
self.reasoning_levels = ["low", "medium", "high"]
self.default_reasoning_level = "low"
def build_single_turn_prompt(
self,
prompt_text: str,
role: str,
content: str,
) -> str:
"""Embed user message into the required prompt template."""
if role == "system":
prompt_text = f"<|start|>system<|message|>{content}<|end|>"
elif role == "assistant_reasoning_effort":
prompt_text = f"<|start|>system<|message|>You are ChatGPT, a large language model trained by OpenAI.\nKnowledge cutoff: 2024-06\nCurrent date: 2025-06-28\n\nReasoning: {content.lower()}\n\n# Valid channels: analysis, commentary, final. Channel must be included for every message.<|end|>"
elif role == "user":
prompt_text += f"<|start|>user<|message|>{content}<|end|>"
elif role == "assistant_analysis":
prompt_text += (
f"<|start|>assistant<|channel|>analysis<|message|>{content}<|end|>"
)
elif role == "assistant_commentary":
prompt_text += (
f"<|start|>assistant<|channel|>commentary<|message|>{content}<|end|>"
)
elif role == "assistant_final":
prompt_text += (
f"<|start|>assistant<|channel|>final<|message|>{content}<|end|>"
)
else:
raise ValueError(f"Unknown role: {role}")
return prompt_text
def parse(
self,
conversation: "Conversation",
max_length: int,
preformatted: bool = False,
train_only_last_turn: bool = False,
) -> List[torch.Tensor]:
# conversation = process_harmony_conversations(conversation)
if not preformatted:
prompt_text = ""
for j, message in enumerate(conversation):
if j == 0 and (
message["role"] != "system"
or message["role"] != "assistant_reasoning_effort"
):
prompt_text = self.build_single_turn_prompt(
prompt_text,
"assistant_reasoning_effort",
self.default_reasoning_level,
)
prompt_text = self.build_single_turn_prompt(
prompt_text, message["role"], message["content"]
)
conversation = prompt_text
if not self.tokenizer.pad_token_id:
self.tokenizer.pad_token_id = self.tokenizer.unk_token_id
encoding = self.tokenizer(
conversation,
return_offsets_mapping=True,
max_length=max_length,
truncation=True,
return_tensors="pt",
add_special_tokens=False,
)
input_ids = encoding.input_ids[0]
offsets = encoding.offset_mapping[0]
loss_mask = torch.zeros(len(input_ids), dtype=torch.long)
# Find spans of assistant responses using regex
# We match `<|start|>assistant` and only extract the content following it.
# This continues until `<|start|>user<|message|>` appears, or until the end of the string.
pattern = re.compile(
r"<\|start\|>assistant([\s\S]*?)(?=<\|start\|>user<\|message\|>|$)"
)
# Find all matching segments
matches = list(pattern.finditer(conversation))
if train_only_last_turn and matches:
matches = [matches[-1]] # Only keep the last match
for match in matches:
# match.start(0) is the start index of the full match (including `<|start|>assistant`)
# match.start(1) is the start index of the first capture group (excluding `<|start|>assistant`)
# match.end(1) is the end index of the content
start_char = match.start(1)
end_char = match.end(1)
# Map character indices to token indices
for idx, (ts, te) in enumerate(offsets):
# Set mask to 1 only if the token's character range falls entirely within the "content area"
if ts >= start_char and te <= end_char:
loss_mask[idx] = 1
return input_ids, loss_mask
class ThinkingParser(GeneralParser):
def __init__(self, tokenizer: PreTrainedTokenizer, chat_template: ChatTemplate):
super().__init__(tokenizer, chat_template)
def apply_chat_template(self, messages, **kwargs) -> str:
if messages[-1]["role"] == "assistant":
conversation_history = self.tokenizer.apply_chat_template(
messages[:-1],
tokenize=False,
add_generation_prompt=True,
add_special_tokens=False,
**kwargs,
)
conversation = (
conversation_history
+ messages[-1]["content"]
+ self.chat_template.end_of_turn_token
)
return conversation
else:
raise Exception(
f"The last message is not assistant but {messages[-1]['role']}"
)
def parse(
self,
conversation: "Conversation",
max_length: int,
preformatted: bool = False,
train_only_last_turn: bool = False,
**kwargs,
) -> Dict[str, List[torch.Tensor]]:
if self.chat_template.enable_thinking:
kwargs["enable_thinking"] = True
else:
pass
return super().parse(
conversation, max_length, preformatted, train_only_last_turn, **kwargs
)
|