Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
@@ -5,6 +5,7 @@ import os
|
|
5 |
import time
|
6 |
from dataclasses import asdict, dataclass
|
7 |
from pathlib import Path
|
|
|
8 |
from types import SimpleNamespace
|
9 |
|
10 |
import gradio as gr
|
@@ -229,7 +230,7 @@ def generate(
|
|
229 |
**asdict(generation_config),
|
230 |
)
|
231 |
|
232 |
-
|
233 |
_ = """full url: https://huggingface.co/TheBloke/mpt-30B-chat-GGML/blob/main/mpt-30b-chat.ggmlv0.q4_1.bin"""
|
234 |
|
235 |
# https://huggingface.co/TheBloke/mpt-30B-chat-GGML
|
@@ -247,15 +248,13 @@ MODEL_FILENAME = "WizardCoder-15B-1.0.ggmlv3.q4_1.bin" # 11.9G
|
|
247 |
|
248 |
# https://huggingface.co/TheBloke/WizardLM-13B-V1.0-Uncensored-GGML
|
249 |
MODEL_FILENAME = "wizardlm-13b-v1.0-uncensored.ggmlv3.q4_1.bin" # 8.4G
|
|
|
250 |
|
251 |
-
|
252 |
-
|
253 |
-
REPO_ID = "TheBloke/
|
254 |
-
if "WizardCoder" in MODEL_FILENAME:
|
255 |
-
REPO_ID = "TheBloke/WizardCoder-15B-1.0-GGML"
|
256 |
|
257 |
-
|
258 |
-
REPO_ID = "TheBloke/WizardLM-13B-V1.0-Uncensored-GGML"
|
259 |
|
260 |
logger.info(f"start dl, {REPO_ID=}, {MODEL_FILENAME=}, {DESTINATION_FOLDER=}")
|
261 |
download_quant(DESTINATION_FOLDER, REPO_ID, MODEL_FILENAME)
|
@@ -365,7 +364,7 @@ with gr.Blocks(
|
|
365 |
# """<center><a href="https://huggingface.co/spaces/mikeee/mpt-30b-chat?duplicate=true"><img src="https://bit.ly/3gLdBN6" alt="Duplicate"></a> and spin a CPU UPGRADE to avoid the queue</center>"""
|
366 |
# )
|
367 |
gr.Markdown(
|
368 |
-
f"""<h4><center>{MODEL_FILENAME}</center></h4>
|
369 |
It takes about 100 seconds for the initial reply
|
370 |
message to appear. Average streaming rate ~1 sec/chat. The bot only speaks English.
|
371 |
|
|
|
5 |
import time
|
6 |
from dataclasses import asdict, dataclass
|
7 |
from pathlib import Path
|
8 |
+
from urllib.parse import urlparse
|
9 |
from types import SimpleNamespace
|
10 |
|
11 |
import gradio as gr
|
|
|
230 |
**asdict(generation_config),
|
231 |
)
|
232 |
|
233 |
+
_ = '''
|
234 |
_ = """full url: https://huggingface.co/TheBloke/mpt-30B-chat-GGML/blob/main/mpt-30b-chat.ggmlv0.q4_1.bin"""
|
235 |
|
236 |
# https://huggingface.co/TheBloke/mpt-30B-chat-GGML
|
|
|
248 |
|
249 |
# https://huggingface.co/TheBloke/WizardLM-13B-V1.0-Uncensored-GGML
|
250 |
MODEL_FILENAME = "wizardlm-13b-v1.0-uncensored.ggmlv3.q4_1.bin" # 8.4G
|
251 |
+
# '''
|
252 |
|
253 |
+
URL = "https://huggingface.co/TheBloke/Wizard-Vicuna-7B-Uncensored-GGML/raw/main/Wizard-Vicuna-7B-Uncensored.ggmlv3.q4_K_M.bin"
|
254 |
+
MODEL_FILENAME = Path(URL).name
|
255 |
+
REPO_ID = "/".join(urlparse(url).path.strip('/').split('/')[:2]) # TheBloke/Wizard-Vicuna-7B-Uncensored-GGML
|
|
|
|
|
256 |
|
257 |
+
DESTINATION_FOLDER = "models"
|
|
|
258 |
|
259 |
logger.info(f"start dl, {REPO_ID=}, {MODEL_FILENAME=}, {DESTINATION_FOLDER=}")
|
260 |
download_quant(DESTINATION_FOLDER, REPO_ID, MODEL_FILENAME)
|
|
|
364 |
# """<center><a href="https://huggingface.co/spaces/mikeee/mpt-30b-chat?duplicate=true"><img src="https://bit.ly/3gLdBN6" alt="Duplicate"></a> and spin a CPU UPGRADE to avoid the queue</center>"""
|
365 |
# )
|
366 |
gr.Markdown(
|
367 |
+
f"""<h4><center>{REPO_ID} {MODEL_FILENAME}</center></h4>
|
368 |
It takes about 100 seconds for the initial reply
|
369 |
message to appear. Average streaming rate ~1 sec/chat. The bot only speaks English.
|
370 |
|