🧠 两小时从 0 训出你的 64M 大模型。Train your own 64M LLM from scratch in 2 hours.
update fast_inference
gongjy7c67ba0b927d
diff --git a/fast_infenence.py b/fast_inference.pysimilarity index 57%rename from fast_infenence.pyrename to fast_inference.pyindex 5ab4eea..ee3a119 100644--- a/fast_infenence.py+++ b/fast_inference.py@@ -1,19 +1,16 @@import json+import random+import numpy as npimport streamlit as stimport torchfrom transformers import AutoModelForCausalLM, AutoTokenizerfrom transformers.generation.utils import GenerationConfig-st.set_page_config(page_title="minimind-v1(108M)")-st.title("minimind-v1(108M)")+st.set_page_config(page_title="MiniMind-V1")+st.title("MiniMind-V1")-model_id = "minimind-v1"+model_id = "./minimind-v1"-# ------------------------------------------------------------------------------temperature = 0.7-top_k = 8-max_seq_len = 1 * 1024-# -----------------------------------------------------------------------------@st.cache_resourcedef load_model_tokenizer():@@ -33,28 +30,41 @@ def load_model_tokenizer():def clear_chat_messages():del st.session_state.messages+ del st.session_state.chat_messagesdef init_chat_messages():with st.chat_message("assistant", avatar='🤖'):- st.markdown("您好,我是由JingyaoGong创造的MiniMind,很高兴为您服务😄")+ st.markdown("我是由JingyaoGong创造的MiniMind,很高兴为您服务😄 \n"+ "注:所有AI生成内容的准确性和立场无法保证,不代表我们的态度或观点。")if "messages" in st.session_state:for message in st.session_state.messages:- avatar = "🧑💻" if message["role"] == "user" else "🤖"+ avatar = "🫡" if message["role"] == "user" else "🤖"with st.chat_message(message["role"], avatar=avatar):st.markdown(message["content"])else:st.session_state.messages = []+ st.session_state.chat_messages = []return st.session_state.messages-# max_new_tokens = st.sidebar.slider("max_new_tokens", 0, 1024, 512, step=1)-# top_p = st.sidebar.slider("top_p", 0.0, 1.0, 0.8, step=0.01)-# top_k = st.sidebar.slider("top_k", 0, 100, 0, step=1)-# temperature = st.sidebar.slider("temperature", 0.0, 2.0, 1.0, step=0.01)-# do_sample = st.sidebar.checkbox("do_sample", value=False)+st.sidebar.title("设定调整")+st.session_state.history_chat_num = st.sidebar.slider("携带历史对话条数", 0, 6, 0, step=2)+st.session_state.max_new_tokens = st.sidebar.slider("最大输入/生成长度", 256, 768, 512, step=1)+st.session_state.top_k = st.sidebar.slider("top_k", 0, 16, 14, step=1)+st.session_state.temperature = st.sidebar.slider("temperature", 0.3, 1.3, 0.5, step=0.01)+++def setup_seed(seed):+ random.seed(seed)+ np.random.seed(seed)+ torch.manual_seed(seed)+ torch.cuda.manual_seed(seed)+ torch.cuda.manual_seed_all(seed)+ torch.backends.cudnn.deterministic = True+ torch.backends.cudnn.benchmark = Falsedef main():@@ -65,32 +75,30 @@ def main():with st.chat_message("user", avatar='🧑💻'):st.markdown(prompt)messages.append({"role": "user", "content": prompt})+ st.session_state.chat_messages.append({"role": "user", "content": '请问,' + prompt + '?'})with st.chat_message("assistant", avatar='🤖'):placeholder = st.empty()+ # Generate a random seed+ random_seed = random.randint(0, 2 ** 32 - 1)+ setup_seed(random_seed)- chat_messages = []- chat_messages.append({"role": "user", "content": '请问,' + prompt})- # print(messages)new_prompt = tokenizer.apply_chat_template(- chat_messages,+ st.session_state.chat_messages[-(st.session_state.history_chat_num + 1):],tokenize=False,add_generation_prompt=True- )[-(max_seq_len - 1):]+ )[-(st.session_state.max_new_tokens - 1):]x = tokenizer(new_prompt).data['input_ids']x = (torch.tensor(x, dtype=torch.long)[None, ...])-- response = ''-with torch.no_grad():- res_y = model.generate(x, tokenizer.eos_token_id, max_new_tokens=max_seq_len, temperature=temperature,- top_k=top_k, stream=True)+ res_y = model.generate(x, tokenizer.eos_token_id, max_new_tokens=st.session_state.max_new_tokens,+ temperature=st.session_state.temperature,+ top_k=st.session_state.top_k, stream=True)try:y = next(res_y)except StopIteration:return- history_idx = 0while y != None:answer = tokenizer.decode(y[0].tolist())if answer and answer[-1] == '�':@@ -99,7 +107,6 @@ def main():except:breakcontinue- # print(answer)if not len(answer):try:y = next(res_y)@@ -107,17 +114,14 @@ def main():breakcontinueplaceholder.markdown(answer)- response = answertry:y = next(res_y)except:break- # if contain_history_chat:- # assistant_answer = answer.replace(new_prompt, "")- # messages.append({"role": "assistant", "content": assistant_answer})-- messages.append({"role": "assistant", "content": response})+ assistant_answer = answer.replace(new_prompt, "")+ messages.append({"role": "assistant", "content": assistant_answer})+ st.session_state.chat_messages.append({"role": "assistant", "content": assistant_answer})st.button("清空对话", on_click=clear_chat_messages)