Spaces:
Runtime error
Runtime error
gen
Browse files
app.py
CHANGED
|
@@ -2,7 +2,6 @@
|
|
| 2 |
import os, logging, torch, streamlit as st
|
| 3 |
from transformers import (
|
| 4 |
AutoTokenizer, AutoModelForCausalLM)
|
| 5 |
-
st.balloons()
|
| 6 |
|
| 7 |
# --------------------- HELPER --------------------- #
|
| 8 |
def C(text, color="yellow"):
|
|
@@ -18,18 +17,31 @@ def C(text, color="yellow"):
|
|
| 18 |
return (
|
| 19 |
f"{color_dict.get(color, None)}"
|
| 20 |
f"{text}{color_dict[None]}")
|
| 21 |
-
st.balloons()
|
| 22 |
|
| 23 |
# ------------------ ENVIORNMENT ------------------- #
|
| 24 |
os.environ["HF_ENDPOINT"] = "https://huggingface.co"
|
| 25 |
device = ("cuda"
|
| 26 |
if torch.cuda.is_available() else "cpu")
|
| 27 |
logging.info(C("[INFO] "f"device = {device}"))
|
| 28 |
-
st.balloons()
|
| 29 |
|
| 30 |
# ------------------ INITITALIZE ------------------- #
|
| 31 |
@st.cache
|
| 32 |
def model_init():
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 33 |
tokenizer = AutoTokenizer.from_pretrained(
|
| 34 |
"ckip-joint/bloom-1b1-zh")
|
| 35 |
model = AutoModelForCausalLM.from_pretrained(
|
|
@@ -44,14 +56,10 @@ def model_init():
|
|
| 44 |
return tokenizer, model
|
| 45 |
|
| 46 |
tokenizer, model = model_init()
|
| 47 |
-
st.balloons()
|
| 48 |
|
| 49 |
try:
|
| 50 |
# ===================== INPUT ====================== #
|
| 51 |
-
# prompt = "\u554F\uFF1A\u53F0\u7063\u6700\u9AD8\u7684\u5EFA\u7BC9\u7269\u662F\uFF1F\u7B54\uFF1A" #@param {type:"string"}
|
| 52 |
prompt = st.text_input("Prompt: ")
|
| 53 |
-
st.balloons()
|
| 54 |
-
|
| 55 |
|
| 56 |
# =================== INFERENCE ==================== #
|
| 57 |
if prompt:
|
|
@@ -59,13 +67,13 @@ try:
|
|
| 59 |
with torch.no_grad():
|
| 60 |
[texts_out] = model.generate(
|
| 61 |
**tokenizer(
|
| 62 |
-
prompt, return_tensors="pt"
|
|
|
|
| 63 |
).to(device))
|
| 64 |
-
st.balloons()
|
| 65 |
output_text = tokenizer.decode(texts_out)
|
| 66 |
st.balloons()
|
| 67 |
st.markdown(output_text)
|
| 68 |
-
|
| 69 |
except Exception as err:
|
| 70 |
st.write(str(err))
|
| 71 |
st.snow()
|
|
|
|
| 2 |
import os, logging, torch, streamlit as st
|
| 3 |
from transformers import (
|
| 4 |
AutoTokenizer, AutoModelForCausalLM)
|
|
|
|
| 5 |
|
| 6 |
# --------------------- HELPER --------------------- #
|
| 7 |
def C(text, color="yellow"):
|
|
|
|
| 17 |
return (
|
| 18 |
f"{color_dict.get(color, None)}"
|
| 19 |
f"{text}{color_dict[None]}")
|
|
|
|
| 20 |
|
| 21 |
# ------------------ ENVIORNMENT ------------------- #
|
| 22 |
os.environ["HF_ENDPOINT"] = "https://huggingface.co"
|
| 23 |
device = ("cuda"
|
| 24 |
if torch.cuda.is_available() else "cpu")
|
| 25 |
logging.info(C("[INFO] "f"device = {device}"))
|
|
|
|
| 26 |
|
| 27 |
# ------------------ INITITALIZE ------------------- #
|
| 28 |
@st.cache
|
| 29 |
def model_init():
|
| 30 |
+
|
| 31 |
+
|
| 32 |
+
|
| 33 |
+
from transformers import GenerationConfig
|
| 34 |
+
|
| 35 |
+
# generation_config, unused_kwargs = GenerationConfig.from_pretrained(
|
| 36 |
+
# "ckip-joint/bloom-1b1-zh",
|
| 37 |
+
# max_new_tokens=200,
|
| 38 |
+
# return_unused_kwargs=True)
|
| 39 |
+
|
| 40 |
+
|
| 41 |
+
|
| 42 |
+
|
| 43 |
+
|
| 44 |
+
|
| 45 |
tokenizer = AutoTokenizer.from_pretrained(
|
| 46 |
"ckip-joint/bloom-1b1-zh")
|
| 47 |
model = AutoModelForCausalLM.from_pretrained(
|
|
|
|
| 56 |
return tokenizer, model
|
| 57 |
|
| 58 |
tokenizer, model = model_init()
|
|
|
|
| 59 |
|
| 60 |
try:
|
| 61 |
# ===================== INPUT ====================== #
|
|
|
|
| 62 |
prompt = st.text_input("Prompt: ")
|
|
|
|
|
|
|
| 63 |
|
| 64 |
# =================== INFERENCE ==================== #
|
| 65 |
if prompt:
|
|
|
|
| 67 |
with torch.no_grad():
|
| 68 |
[texts_out] = model.generate(
|
| 69 |
**tokenizer(
|
| 70 |
+
prompt, return_tensors="pt",
|
| 71 |
+
max_new_tokens=200,
|
| 72 |
).to(device))
|
|
|
|
| 73 |
output_text = tokenizer.decode(texts_out)
|
| 74 |
st.balloons()
|
| 75 |
st.markdown(output_text)
|
| 76 |
+
|
| 77 |
except Exception as err:
|
| 78 |
st.write(str(err))
|
| 79 |
st.snow()
|