This commit is contained in:
1
patches/architectures/aliases.json
Normal file
1
patches/architectures/aliases.json
Normal file
@@ -0,0 +1 @@
|
||||
{}
|
||||
13
patches/architectures/whitelist.txt
Normal file
13
patches/architectures/whitelist.txt
Normal file
@@ -0,0 +1,13 @@
|
||||
# 本仓本地兼容白名单,不是平台公布的完整列表。
|
||||
# 只收录 vLLM 0.17 文本生成里常见、且本引擎准备接受的架构名。
|
||||
LlamaForCausalLM
|
||||
Qwen2ForCausalLM
|
||||
Qwen3ForCausalLM
|
||||
Qwen3MoeForCausalLM
|
||||
MistralForCausalLM
|
||||
MixtralForCausalLM
|
||||
Gemma2ForCausalLM
|
||||
Phi3ForCausalLM
|
||||
InternLM2ForCausalLM
|
||||
BaichuanForCausalLM
|
||||
ChatGLMForConditionalGeneration
|
||||
1
patches/chat_template/default_chat_template.jinja
Normal file
1
patches/chat_template/default_chat_template.jinja
Normal file
@@ -0,0 +1 @@
|
||||
{% for message in messages %}{{ '<|' + message['role'] + '|>\n' + message['content'] + '<|end|>\n' }}{% endfor %}{% if add_generation_prompt %}{{ '<|assistant|>\n' }}{% endif %}
|
||||
5
patches/tokenizer/README.md
Normal file
5
patches/tokenizer/README.md
Normal file
@@ -0,0 +1,5 @@
|
||||
# tokenizer 修补(I5,未开启)
|
||||
|
||||
现有 bi150 引擎已经覆盖 tokenizer。本仓先补 chat_template(R2)和 architecture 白名单(R4)。
|
||||
|
||||
`APPLY_TOKENIZER_PATCH` 默认为 `0`。在确认某类模型只靠 R2/R4 仍无法加载之前,不要打开这项。
|
||||
Reference in New Issue
Block a user