tweak fastchat conversation with a monkeypatch to get individual turns

src/axolotl/monkeypatch/fastchat_conversation_turns.py

@@ -0,0 +1,26 @@
+"""
+monkeypatch to add a get_turns method
+"""
+
+import logging
+from typing import Generator, Tuple
+
+LOG = logging.getLogger("axolotl.monkeypatch.fastchat_conversation_turns")
+
+
+def get_turns(self) -> Generator[Tuple[str, str], None, None]:
+    # seps = [self.sep, self.sep2]
+    preamble = self.system_message + self.sep
+    yield ("SYSTEM:", preamble)
+    for _, (role, message) in enumerate(self.messages):
+        if message:
+            yield (role + ":", " " + message)
+        else:
+            LOG.warning(f"role with empty message: {role}")
+            yield (role + ":", "")
+
+
+def add_get_turns_to_conversation():
+    import fastchat.conversation
+
+    fastchat.conversation.Conversation.get_turns = get_turns

src/axolotl/prompt_tokenizers.py

@@ -9,6 +9,9 @@
 from fastchat.conversation import Conversation
 from transformers import BatchEncoding, PreTrainedTokenizer
 
+from axolotl.monkeypatch.fastchat_conversation_turns import (
+    add_get_turns_to_conversation,
+)
 from axolotl.prompters import IGNORE_TOKEN_ID
 
 LOG = logging.getLogger("axolotl")
@@ -19,6 +22,8 @@
 LLAMA_DEFAULT_BOS_TOKEN = "<s>"  # nosec
 LLAMA_DEFAULT_UNK_TOKEN = "<unk>"  # nosec
 
+add_get_turns_to_conversation()
+
 
 class InvalidDataException(Exception):
     """
@@ -361,7 +366,7 @@ def tokenize_prompt(self, prompt):
                 self.prompter.build_prompt(self.get_conversation_thread(prompt))
             ):
                 if isinstance(part, tuple):
-                    if part[0] == conversation.roles[0]:
+                    if part[0] == conversation.roles[0] + ":":
                         turn = part[0] + part[1] if not user_token else part[1]
                         # this is still the user query, we should
                         if not part[1].strip():
@@ -375,7 +380,7 @@ def tokenize_prompt(self, prompt):
                             res["input_ids"] = [user_token, *res["input_ids"]]
                         # everything from this is masked out from the labels
                         labels = [IGNORE_TOKEN_ID] * len(res["input_ids"])
-                    elif part[0] == conversation.roles[1]:
+                    elif part[0] == conversation.roles[1] + ":":
                         # TODO label assistant token/tokens w/ IGNORE_TOKEN_ID
                         turn = part[0] + part[1] if not assistant_token else part[1]
                         # this should be the assistant response, should end with an eos token
@@ -403,6 +408,7 @@ def tokenize_prompt(self, prompt):
                         labels = [IGNORE_TOKEN_ID] * len(res["input_ids"])
                     else:
                         LOG.warning(f"unhandled role: {part[0]}")
+                        continue
 
                 # pylint: disable=duplicate-code
                 result, current_len = parse_tokenized_to_result(

src/axolotl/prompters.py

@@ -289,5 +289,5 @@ def build_prompt(self, source) -> Generator[str, None, None]:
             assert role == conv.roles[j % 2], SHAREGPT_ASSERTION_FAILED_ROLE
             conv.append_message(role, sentence["value"])
 
-        for part in conv.get_prompt():
+        for part in conv.get_turns():
             yield part