Skip to content

convert: ZAYA special tokens are CONTROL, so chat templates tokenize #27

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
bong-water-water-bong merged 1 commit into from
Sep 27, 2026
Merged

convert: ZAYA special tokens are CONTROL, so chat templates tokenize #27

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
22 changes: 20 additions & 2 deletions conversion/zaya.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,6 +31,24 @@
from .qwenvl import Qwen2VLVisionModel



def _special_tokens_as_control(tokenizer_dir, toktypes):
"""Mark every token tokenizer.json lists as special: true as CONTROL.

LlamaHfVocab leaves some of them NORMAL (ZAYA1's <|im_start|> and <eos>, which are also in the
base vocab), and llama.cpp matches special-token text in prompts only for CONTROL and
USER_DEFINED tokens, so the chat template's <|im_start|> reached the model spelled out as text."""
import json
from pathlib import Path
path = Path(tokenizer_dir) / "tokenizer.json"
if not path.is_file():
return toktypes
for added in json.loads(path.read_text(encoding="utf-8")).get("added_tokens", []):
tid = added.get("id")
if added.get("special") and isinstance(tid, int) and 0 <= tid < len(toktypes):
toktypes[tid] = gguf.TokenType.CONTROL
return toktypes

@ModelBase.register("ZayaForCausalLM")
class ZayaModel(TextModel):
"""Zyphra ZAYA1 (transformers naming): every layer is CCA attention + a top-1 MoE."""
Expand Down Expand Up @@ -260,7 +278,7 @@ def set_vocab(self):
self.gguf_writer.add_tokenizer_model("gemma4")
self.gguf_writer.add_token_list(tokens)
self.gguf_writer.add_token_scores(scores)
self.gguf_writer.add_token_types(toktypes)
self.gguf_writer.add_token_types(_special_tokens_as_control(self._tokenizer_dir(), toktypes))

special_vocab = gguf.SpecialVocab(self.dir_model, load_merges=True)
special_vocab.add_to_gguf(self.gguf_writer)
Expand Down Expand Up @@ -386,7 +404,7 @@ def set_vocab(self):
self.gguf_writer.add_tokenizer_model("gemma4")
self.gguf_writer.add_token_list(tokens)
self.gguf_writer.add_token_scores(scores)
self.gguf_writer.add_token_types(toktypes)
self.gguf_writer.add_token_types(_special_tokens_as_control(self._tokenizer_dir(), toktypes))

special_vocab = gguf.SpecialVocab(self.dir_model, load_merges=True)
special_vocab.chat_template = self._CHAT_TEMPLATE
Expand Down
Loading