happyz
/
llama.cpp
mirror of https://github.com/ggerganov/llama.cpp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Merge e62f6849bd into 2ba9adc093

This commit is contained in:
DAN™ 2026-02-16 15:53:48 +02:00 committed by GitHub
parent 2ba9adc093 e62f6849bd
commit 1e1ce8b1f3
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
4 changed files with 15 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
convert_hf_to_gguf.py
View File

@ -1049,6 +1049,9 @@ class TextModel(ModelBase):
if chkhsh == "9ca2dd618e8afaf09731a7cf6e2105b373ba6a1821559f258b272fe83e6eb902":
# ref: https://huggingface.co/zai-org/GLM-4.5-Air
res = "glm4"
if chkhsh == "cdf5f35325780597efd76153d4d1c16778f766173908894c04afc20108536267":
# ref: https://huggingface.co/zai-org/GLM-4.7-Flash
res = "glm4"
if chkhsh == "1431a23e583c97432bc230bff598d103ddb5a1f89960c8f1d1051aaa944d0b35":
# ref: https://huggingface.co/sapienzanlp/Minerva-7B-base-v1.0
res = "minerva-7b"
@ -1082,9 +1085,6 @@ class TextModel(ModelBase):
if chkhsh == "b3d1dd861f1d4c5c0d2569ce36baf3f90fe8a102db3de50dd71ff860d91be3df":
# ref: https://huggingface.co/aari1995/German_Semantic_V3
res = "jina-v2-de"
if chkhsh == "cdf5f35325780597efd76153d4d1c16778f766173908894c04afc20108536267":
# ref: https://huggingface.co/zai-org/GLM-4.7-Flash
res = "glm4"
if chkhsh == "0ef9807a4087ebef797fc749390439009c3b9eda9ad1a097abbe738f486c01e5":
# ref: https://huggingface.co/meta-llama/Meta-Llama-3-8B
res = "llama-bpe"
@ -1265,6 +1265,9 @@ class TextModel(ModelBase):
if chkhsh == "d30d75d9059f1aa2c19359de71047b3ae408c70875e8a3ccf8c5fba56c9d8af4":
# ref: https://huggingface.co/Qwen/Qwen3.5-9B-Instruct
res = "qwen35"
if chkhsh == "b4b8ca1f9769494fbd956ebc4c249de6131fb277a4a3345a7a92c7dd7a55808d":
# ref: https://huggingface.co/jdopensource/JoyAI-LLM-Flash
res = "joyai-llm"
if res is None:
logger.warning("\n")

5
convert_hf_to_gguf_update.py
View File

@ -148,7 +148,8 @@ models = [
{"name": "youtu", "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/tencent/Youtu-LLM-2B", },
{"name": "solar-open", "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/upstage/Solar-Open-100B", },
{"name": "exaone-moe", "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/LGAI-EXAONE/K-EXAONE-236B-A23B", },
{"name": "qwen35", "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/Qwen/Qwen3.5-9B-Instruct", }
{"name": "qwen35", "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/Qwen/Qwen3.5-9B-Instruct", },
{"name": "joyai-llm", "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/jdopensource/JoyAI-LLM-Flash", },
]
# some models are known to be broken upstream, so we will skip them as exceptions
@ -158,6 +159,7 @@ pre_computed_hashes = [
{"name": "chatglm-bpe", "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/THUDM/glm-4-9b-chat", "chkhsh": "81d72c7348a9f0ebe86f23298d37debe0a5e71149e29bd283904c02262b27516"},
{"name": "glm4", "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/THUDM/glm-4-9b-hf", "chkhsh": "a1336059768a55c99a734006ffb02203cd450fed003e9a71886c88acf24fdbc2"},
{"name": "glm4", "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/zai-org/GLM-4.5-Air", "chkhsh": "9ca2dd618e8afaf09731a7cf6e2105b373ba6a1821559f258b272fe83e6eb902"},
{"name": "glm4", "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/zai-org/GLM-4.7-Flash", "chkhsh": "cdf5f35325780597efd76153d4d1c16778f766173908894c04afc20108536267"},
{"name": "minerva-7b", "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/sapienzanlp/Minerva-7B-base-v1.0", "chkhsh": "1431a23e583c97432bc230bff598d103ddb5a1f89960c8f1d1051aaa944d0b35"},
{"name": "hunyuan", "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/tencent/Hunyuan-A13B-Instruct", "chkhsh": "7e57df22b1fe23a7b1e1c7f3dc4e3f96d43a4eb0836d0c6bdc3436d7b2f1c664"},
{"name": "hunyuan-dense", "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/tencent/Hunyuan-4B-Instruct", "chkhsh": "bba3b3366b646dbdded5dbc42d59598b849371afc42f7beafa914afaa5b70aa6"},
@ -171,7 +173,6 @@ pre_computed_hashes = [
{"name": "grok-2", "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/alvarobartt/grok-2-tokenizer", "chkhsh": "66b8d4e19ab16c3bfd89bce5d785fb7e0155e8648708a1f42077cb9fe002c273"},
# jina-v2-de variants
{"name": "jina-v2-de", "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/aari1995/German_Semantic_V3", "chkhsh": "b3d1dd861f1d4c5c0d2569ce36baf3f90fe8a102db3de50dd71ff860d91be3df"},
{"name": "glm4", "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/zai-org/GLM-4.7-Flash", "chkhsh": "cdf5f35325780597efd76153d4d1c16778f766173908894c04afc20108536267"},
]

5
src/llama-vocab.cpp
View File

@ -308,6 +308,7 @@ struct llm_tokenizer_bpe : llm_tokenizer {
break;
case LLAMA_VOCAB_PRE_TYPE_DEEPSEEK3_LLM:
case LLAMA_VOCAB_PRE_TYPE_HUNYUAN_DENSE:
case LLAMA_VOCAB_PRE_TYPE_JOYAI_LLM:
regex_exprs = {
"\\p{N}{1,3}",
"[一-龥぀-ゟ゠-ヿ]+",
@ -2039,6 +2040,10 @@ void llama_vocab::impl::load(llama_model_loader & ml, const LLM_KV & kv) {
tokenizer_pre == "hunyuan-dense") {
pre_type = LLAMA_VOCAB_PRE_TYPE_HUNYUAN_DENSE;
clean_spaces = false;
} else if (
tokenizer_pre == "joyai-llm") {
pre_type = LLAMA_VOCAB_PRE_TYPE_JOYAI_LLM;
clean_spaces = false;
} else if (
tokenizer_pre == "kimi-k2") {
pre_type = LLAMA_VOCAB_PRE_TYPE_KIMI_K2;

1
src/llama-vocab.h
View File

@ -55,6 +55,7 @@ enum llama_vocab_pre_type {
LLAMA_VOCAB_PRE_TYPE_YOUTU = 44,
LLAMA_VOCAB_PRE_TYPE_EXAONE_MOE = 45,
LLAMA_VOCAB_PRE_TYPE_QWEN35 = 46,
LLAMA_VOCAB_PRE_TYPE_JOYAI_LLM = 47,
};
struct LLM_KV;