diff --git a/examples/fastapi_server.py b/examples/fastapi_server.py index e601f0f..93b6edb 100644 --- a/examples/fastapi_server.py +++ b/examples/fastapi_server.py @@ -5,9 +5,11 @@ from llama_cpp import Llama from fastapi import FastAPI from pydantic import BaseModel, BaseSettings, Field + class Settings(BaseSettings): model: str + app = FastAPI( title="🦙 llama.cpp Python API", version="0.0.1", @@ -15,6 +17,7 @@ app = FastAPI( settings = Settings() llama = Llama(settings.model) + class CompletionRequest(BaseModel): prompt: str suffix: Optional[str] = Field(None) @@ -31,12 +34,11 @@ class CompletionRequest(BaseModel): schema_extra = { "example": { "prompt": "\n\n### Instructions:\nWhat is the capital of France?\n\n### Response:\n", - "stop": ["\n", "###"] + "stop": ["\n", "###"], } } - @app.post("/v1/completions") def completions(request: CompletionRequest): - return llama(**request.dict()) \ No newline at end of file + return llama(**request.dict()) diff --git a/examples/high_level_api_basic_inference.py b/examples/high_level_api_basic_inference.py index f6f36d2..a72adf1 100644 --- a/examples/high_level_api_basic_inference.py +++ b/examples/high_level_api_basic_inference.py @@ -9,6 +9,11 @@ args = parser.parse_args() llm = Llama(model_path=args.model) -output = llm("Question: What are the names of the planets in the solar system? Answer: ", max_tokens=48, stop=["Q:", "\n"], echo=True) +output = llm( + "Question: What are the names of the planets in the solar system? Answer: ", + max_tokens=48, + stop=["Q:", "\n"], + echo=True, +) -print(json.dumps(output, indent=2)) \ No newline at end of file +print(json.dumps(output, indent=2)) diff --git a/examples/langchain_custom_llm.py b/examples/langchain_custom_llm.py index 5d4806d..6ffd78e 100644 --- a/examples/langchain_custom_llm.py +++ b/examples/langchain_custom_llm.py @@ -5,6 +5,7 @@ from llama_cpp import Llama from langchain.llms.base import LLM from typing import Optional, List, Mapping, Any + class LlamaLLM(LLM): model_path: str llm: Llama @@ -16,7 +17,7 @@ class LlamaLLM(LLM): def __init__(self, model_path: str, **kwargs: Any): model_path = model_path llm = Llama(model_path=model_path) - super().__init__(model_path=model_path, llm=llm, **kwargs) + super().__init__(model_path=model_path, llm=llm, **kwargs) def _call(self, prompt: str, stop: Optional[List[str]] = None) -> str: response = self.llm(prompt, stop=stop or []) @@ -26,6 +27,7 @@ class LlamaLLM(LLM): def _identifying_params(self) -> Mapping[str, Any]: return {"model_path": self.model_path} + parser = argparse.ArgumentParser() parser.add_argument("-m", "--model", type=str, default="./models/...") args = parser.parse_args() @@ -34,7 +36,9 @@ args = parser.parse_args() llm = LlamaLLM(model_path=args.model) # Basic Q&A -answer = llm("Question: What is the capital of France? Answer: ", stop=["Question:", "\n"]) +answer = llm( + "Question: What is the capital of France? Answer: ", stop=["Question:", "\n"] +) print(f"Answer: {answer.strip()}") # Using in a chain @@ -48,4 +52,4 @@ prompt = PromptTemplate( chain = LLMChain(llm=llm, prompt=prompt) # Run the chain only specifying the input variable. -print(chain.run("colorful socks")) \ No newline at end of file +print(chain.run("colorful socks")) diff --git a/examples/low_level_api_inference.py b/examples/low_level_api_inference.py index 3e8d9a9..0031f2e 100644 --- a/examples/low_level_api_inference.py +++ b/examples/low_level_api_inference.py @@ -27,7 +27,15 @@ embd = embd_inp n = 8 for i in range(n): - id = llama_cpp.llama_sample_top_p_top_k(ctx, (llama_cpp.c_int * len(embd))(*embd), n_of_tok + i, 40, 0.8, 0.2, 1.0/0.85) + id = llama_cpp.llama_sample_top_p_top_k( + ctx, + (llama_cpp.c_int * len(embd))(*embd), + n_of_tok + i, + 40, + 0.8, + 0.2, + 1.0 / 0.85, + ) embd.append(id) @@ -38,4 +46,4 @@ for i in range(n): llama_cpp.llama_free(ctx) -print(prediction.decode("utf-8")) \ No newline at end of file +print(prediction.decode("utf-8")) diff --git a/llama_cpp/llama.py b/llama_cpp/llama.py index f46d741..7d67646 100644 --- a/llama_cpp/llama.py +++ b/llama_cpp/llama.py @@ -5,6 +5,7 @@ from typing import List, Optional from . import llama_cpp + class Llama: def __init__( self, @@ -82,7 +83,10 @@ class Llama: for i in range(max_tokens): tokens_seen = prompt_tokens + completion_tokens - last_n_tokens = [0] * max(0, self.last_n - tokens_seen) + [self.tokens[j] for j in range(max(tokens_seen - self.last_n, 0), tokens_seen)] + last_n_tokens = [0] * max(0, self.last_n - tokens_seen) + [ + self.tokens[j] + for j in range(max(tokens_seen - self.last_n, 0), tokens_seen) + ] token = llama_cpp.llama_sample_top_p_top_k( self.ctx, @@ -128,9 +132,8 @@ class Llama: self.ctx, )[:logprobs] - return { - "id": f"cmpl-{str(uuid.uuid4())}", # Likely to change + "id": f"cmpl-{str(uuid.uuid4())}", # Likely to change "object": "text_completion", "created": int(time.time()), "model": self.model_path, @@ -151,5 +154,3 @@ class Llama: def __del__(self): llama_cpp.llama_free(self.ctx) - - diff --git a/llama_cpp/llama_cpp.py b/llama_cpp/llama_cpp.py index e53c704..0947187 100644 --- a/llama_cpp/llama_cpp.py +++ b/llama_cpp/llama_cpp.py @@ -1,6 +1,15 @@ import ctypes -from ctypes import c_int, c_float, c_double, c_char_p, c_void_p, c_bool, POINTER, Structure +from ctypes import ( + c_int, + c_float, + c_double, + c_char_p, + c_void_p, + c_bool, + POINTER, + Structure, +) import pathlib @@ -13,26 +22,32 @@ lib = ctypes.CDLL(str(libfile)) llama_token = c_int llama_token_p = POINTER(llama_token) + class llama_token_data(Structure): _fields_ = [ - ('id', llama_token), # token id - ('p', c_float), # probability of the token - ('plog', c_float), # log probability of the token + ("id", llama_token), # token id + ("p", c_float), # probability of the token + ("plog", c_float), # log probability of the token ] + llama_token_data_p = POINTER(llama_token_data) + class llama_context_params(Structure): _fields_ = [ - ('n_ctx', c_int), # text context - ('n_parts', c_int), # -1 for default - ('seed', c_int), # RNG seed, 0 for random - ('f16_kv', c_bool), # use fp16 for KV cache - ('logits_all', c_bool), # the llama_eval() call computes all logits, not just the last one - - ('vocab_only', c_bool), # only load the vocabulary, no weights + ("n_ctx", c_int), # text context + ("n_parts", c_int), # -1 for default + ("seed", c_int), # RNG seed, 0 for random + ("f16_kv", c_bool), # use fp16 for KV cache + ( + "logits_all", + c_bool, + ), # the llama_eval() call computes all logits, not just the last one + ("vocab_only", c_bool), # only load the vocabulary, no weights ] + llama_context_params_p = POINTER(llama_context_params) llama_context_p = c_void_p @@ -74,7 +89,15 @@ lib.llama_token_bos.restype = llama_token lib.llama_token_eos.argtypes = [] lib.llama_token_eos.restype = llama_token -lib.llama_sample_top_p_top_k.argtypes = [llama_context_p, llama_token_p, c_int, c_int, c_double, c_double, c_double] +lib.llama_sample_top_p_top_k.argtypes = [ + llama_context_p, + llama_token_p, + c_int, + c_int, + c_double, + c_double, + c_double, +] lib.llama_sample_top_p_top_k.restype = llama_token lib.llama_print_timings.argtypes = [llama_context_p] @@ -86,45 +109,71 @@ lib.llama_reset_timings.restype = None lib.llama_print_system_info.argtypes = [] lib.llama_print_system_info.restype = c_char_p + # Python functions def llama_context_default_params() -> llama_context_params: params = lib.llama_context_default_params() return params -def llama_init_from_file(path_model: bytes, params: llama_context_params) -> llama_context_p: + +def llama_init_from_file( + path_model: bytes, params: llama_context_params +) -> llama_context_p: """Various functions for loading a ggml llama model. Allocate (almost) all memory needed for the model. - Return NULL on failure """ + Return NULL on failure""" return lib.llama_init_from_file(path_model, params) + def llama_free(ctx: llama_context_p): """Free all allocated memory""" lib.llama_free(ctx) -def llama_model_quantize(fname_inp: bytes, fname_out: bytes, itype: c_int, qk: c_int) -> c_int: + +def llama_model_quantize( + fname_inp: bytes, fname_out: bytes, itype: c_int, qk: c_int +) -> c_int: """Returns 0 on success""" return lib.llama_model_quantize(fname_inp, fname_out, itype, qk) -def llama_eval(ctx: llama_context_p, tokens: llama_token_p, n_tokens: c_int, n_past: c_int, n_threads: c_int) -> c_int: + +def llama_eval( + ctx: llama_context_p, + tokens: llama_token_p, + n_tokens: c_int, + n_past: c_int, + n_threads: c_int, +) -> c_int: """Run the llama inference to obtain the logits and probabilities for the next token. tokens + n_tokens is the provided batch of new tokens to process n_past is the number of tokens to use from previous eval calls Returns 0 on success""" return lib.llama_eval(ctx, tokens, n_tokens, n_past, n_threads) -def llama_tokenize(ctx: llama_context_p, text: bytes, tokens: llama_token_p, n_max_tokens: c_int, add_bos: c_bool) -> c_int: + +def llama_tokenize( + ctx: llama_context_p, + text: bytes, + tokens: llama_token_p, + n_max_tokens: c_int, + add_bos: c_bool, +) -> c_int: """Convert the provided text into tokens. The tokens pointer must be large enough to hold the resulting tokens. Returns the number of tokens on success, no more than n_max_tokens - Returns a negative number on failure - the number of tokens that would have been returned""" + Returns a negative number on failure - the number of tokens that would have been returned + """ return lib.llama_tokenize(ctx, text, tokens, n_max_tokens, add_bos) + def llama_n_vocab(ctx: llama_context_p) -> c_int: return lib.llama_n_vocab(ctx) + def llama_n_ctx(ctx: llama_context_p) -> c_int: return lib.llama_n_ctx(ctx) + def llama_get_logits(ctx: llama_context_p): """Token logits obtained from the last call to llama_eval() The logits for the last token are stored in the last row @@ -133,25 +182,42 @@ def llama_get_logits(ctx: llama_context_p): Cols: n_vocab""" return lib.llama_get_logits(ctx) + def llama_token_to_str(ctx: llama_context_p, token: int) -> bytes: """Token Id -> String. Uses the vocabulary in the provided context""" return lib.llama_token_to_str(ctx, token) + def llama_token_bos() -> llama_token: return lib.llama_token_bos() + def llama_token_eos() -> llama_token: return lib.llama_token_eos() -def llama_sample_top_p_top_k(ctx: llama_context_p, last_n_tokens_data: llama_token_p, last_n_tokens_size: c_int, top_k: c_int, top_p: c_double, temp: c_double, repeat_penalty: c_double) -> llama_token: - return lib.llama_sample_top_p_top_k(ctx, last_n_tokens_data, last_n_tokens_size, top_k, top_p, temp, repeat_penalty) + +def llama_sample_top_p_top_k( + ctx: llama_context_p, + last_n_tokens_data: llama_token_p, + last_n_tokens_size: c_int, + top_k: c_int, + top_p: c_double, + temp: c_double, + repeat_penalty: c_double, +) -> llama_token: + return lib.llama_sample_top_p_top_k( + ctx, last_n_tokens_data, last_n_tokens_size, top_k, top_p, temp, repeat_penalty + ) + def llama_print_timings(ctx: llama_context_p): lib.llama_print_timings(ctx) + def llama_reset_timings(ctx: llama_context_p): lib.llama_reset_timings(ctx) + def llama_print_system_info() -> bytes: """Print system informaiton""" return lib.llama_print_system_info()