CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-10-27 08:21:30 +00:00
Code Issues Packages Projects Releases Wiki Activity

model : add Kimi-K2 support (#14654)

Browse Source 
* Kimi-K2 conversion

* add Kimi_K2  pre type

* Kimi-K2

* Kimi-K2 unicode

* Kimi-K2

* LLAMA_MAX_EXPERTS 384

* fix vocab iteration

* regex space fix

* add kimi-k2 to pre_computed_hashes

* Updated with kimi-k2 get_vocab_base_pre hash

* fix whitespaces

* fix flake errors

* remove more unicode.cpp whitespaces

* change set_vocab() flow

* add moonshotai-Kimi-K2.jinja to /models/templates/

* update moonshotai-Kimi-K2.jinja

* add kimi-k2 chat template

* add kimi-k2

* update NotImplementedError

Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com>

* except Exception

Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com>

* LLM_CHAT_TEMPLATE_KIMI_K2 if(add_ass){}

---------

Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com>
This commit is contained in:
Gabriel Larson
2025-07-15 14:54:22 -05:00
committed by GitHub
parent ba1ceb3456
commit 4a4f426944
10 changed files with 345 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

56
convert_hf_to_gguf.py
View File

@@ -840,6 +840,9 @@ class TextModel(ModelBase):
if chkhsh == "169bf0296a13c4d9b7672313f749eb36501d931022de052aad6e36f2bf34dd51":
# ref: https://huggingface.co/LiquidAI/LFM2-Tokenizer
res = "lfm2"
if chkhsh == "81212dc7cdb7e0c1074ca62c5aeab0d43c9f52b8a737be7b12a777c953027890":
# ref: https://huggingface.co/moonshotai/Kimi-K2-Base
res = "kimi-k2"
if res is None:
logger.warning("\n")
@@ -5739,7 +5742,58 @@ class DeepseekV2Model(TextModel):
model_arch = gguf.MODEL_ARCH.DEEPSEEK2
def set_vocab(self):
self._set_vocab_gpt2()
try:
self._set_vocab_gpt2()
return
except Exception:
pass
from transformers import AutoTokenizer
tokenizer = AutoTokenizer.from_pretrained(self.dir_model, trust_remote_code=True)
tokpre = self.get_vocab_base_pre(tokenizer)
if tokpre == "kimi-k2":
# Build merges list using the approach similar to HunYuanMoE
merges = []
vocab = {}
mergeable_ranks = tokenizer.model._mergeable_ranks
for token, rank in mergeable_ranks.items():
vocab[QwenModel.token_bytes_to_string(token)] = rank
if len(token) == 1:
continue
merged = QwenModel.bpe(mergeable_ranks, token, max_rank=rank)
if len(merged) == 2:
merges.append(' '.join(map(QwenModel.token_bytes_to_string, merged)))
# Build token list
vocab_size = self.hparams["vocab_size"]
special_tokens = tokenizer.special_tokens
reverse_vocab = {id_ : encoded_tok for encoded_tok, id_ in {**vocab, **special_tokens}.items()}
tokens: list[str] = []
toktypes: list[int] = []
for i in range(vocab_size):
if i not in reverse_vocab:
tokens.append(f"[PAD{i}]")
toktypes.append(gguf.TokenType.UNUSED)
else:
token = reverse_vocab[i]
tokens.append(token)
if i in special_tokens.values():
toktypes.append(gguf.TokenType.CONTROL)
else:
toktypes.append(gguf.TokenType.NORMAL)
self.gguf_writer.add_tokenizer_model("gpt2")
self.gguf_writer.add_tokenizer_pre(tokpre)
self.gguf_writer.add_token_list(tokens)
self.gguf_writer.add_token_types(toktypes)
self.gguf_writer.add_token_merges(merges)
special_vocab = gguf.SpecialVocab(self.dir_model, load_merges=False)
special_vocab.add_to_gguf(self.gguf_writer)
else:
raise NotImplementedError(f"Deepseek pre-tokenizer {tokpre!r} is not supported yet!")
def set_gguf_parameters(self):