Open uniartisan opened 1 month ago
注释/代码 (Comments/Code)
文档 (Documentation)
懒人函数 (Convenience Functions) 功能类函数 (Functional Functions) 训练函数 training() V1 版本暂时不实现 (Training function training() not implemented in V1)
training()
命令行聊天函数 cmd_chat(state=None, parameters) (Command-line Chat Function)
cmd_chat(state=None, parameters)
聊天续写函数 chat(context, state=None, parameters) (Chat Continuation Function)
chat(context, state=None, parameters)
续写函数 completion(prompt, state=None, parameters) (Completion Function)
completion(prompt, state=None, parameters)
状态获取函数 state() (State Retrieval Function)
state()
配置类函数 (Configuration Functions) 分词器配置函数 tokenizer("rwkv_vocab_v20230424.txt") (Tokenizer Configuration Function)
tokenizer("rwkv_vocab_v20230424.txt")
策略配置函数 RWKV(strategy=None,layer) (Strategy Configuration Function)
RWKV(strategy=None,layer)
strategy
State 文件加载函数 load_state("state_file_path" or list [Tensor] ) (State File Loading Function)
load_state("state_file_path" or list [Tensor] )
Others
发布 PIP 包 (Publish PIP Package)
后面有空了可以迁移到pyproject,toml
注释/代码 (Comments/Code)
文档 (Documentation)
懒人函数 (Convenience Functions) 功能类函数 (Functional Functions)
训练函数training()
V1 版本暂时不实现 (Training functiontraining()
not implemented in V1)用 accerlate 写一个分布式训练(Write a distributed training using accelerate) @uniartisan命令行聊天函数
cmd_chat(state=None, parameters)
(Command-line Chat Function)聊天续写函数
chat(context, state=None, parameters)
(Chat Continuation Function)续写函数
completion(prompt, state=None, parameters)
(Completion Function)状态获取函数
state()
(State Retrieval Function)配置类函数 (Configuration Functions) 分词器配置函数
tokenizer("rwkv_vocab_v20230424.txt")
(Tokenizer Configuration Function)策略配置函数
RWKV(strategy=None,layer)
(Strategy Configuration Function)strategy
(fp16/int8/ nf4 ,"layer" = 30) (Support customstrategy
(fp16/int8/nf4, "layer" = 30))State 文件加载函数
load_state("state_file_path" or list [Tensor] )
(State File Loading Function)Others
发布 PIP 包 (Publish PIP Package)