Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -2,17 +2,21 @@ import os
|
|
| 2 |
import gradio as gr
|
| 3 |
import copy
|
| 4 |
from llama_cpp import Llama
|
| 5 |
-
from huggingface_hub import hf_hub_download
|
| 6 |
|
| 7 |
|
| 8 |
-
|
| 9 |
-
|
| 10 |
-
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
|
| 15 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 16 |
|
| 17 |
|
| 18 |
def generate_text(
|
|
@@ -26,7 +30,13 @@ def generate_text(
|
|
| 26 |
temp = ""
|
| 27 |
input_prompt = f"[INST] <<SYS>>\n{system_message}\n<</SYS>>\n\n "
|
| 28 |
for interaction in history:
|
| 29 |
-
input_prompt =
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 30 |
|
| 31 |
input_prompt = input_prompt + str(message) + " [/INST] "
|
| 32 |
|
|
@@ -58,11 +68,11 @@ demo = gr.ChatInterface(
|
|
| 58 |
title="llama-cpp-python on GPU",
|
| 59 |
description="Running LLM with https://github.com/abetlen/llama-cpp-python",
|
| 60 |
examples=[
|
| 61 |
-
[
|
| 62 |
-
[
|
| 63 |
-
[
|
| 64 |
-
[
|
| 65 |
-
[
|
| 66 |
],
|
| 67 |
cache_examples=False,
|
| 68 |
retry_btn=None,
|
|
|
|
| 2 |
import gradio as gr
|
| 3 |
import copy
|
| 4 |
from llama_cpp import Llama
|
| 5 |
+
from huggingface_hub import hf_hub_download
|
| 6 |
|
| 7 |
|
| 8 |
+
try:
|
| 9 |
+
llm = Llama(
|
| 10 |
+
model_path=hf_hub_download(
|
| 11 |
+
repo_id=os.environ.get("REPO_ID", "microsoft/Phi-3-mini-4k-instruct-gguf"),
|
| 12 |
+
filename=os.environ.get("MODEL_FILE", "Phi-3-mini-4k-instruct-q4.gguf"),
|
| 13 |
+
),
|
| 14 |
+
n_ctx=2048,
|
| 15 |
+
n_gpu_layers=-1, # change n_gpu_layers if you have more or less VRAM
|
| 16 |
+
)
|
| 17 |
+
|
| 18 |
+
except Exception as e:
|
| 19 |
+
print(e)
|
| 20 |
|
| 21 |
|
| 22 |
def generate_text(
|
|
|
|
| 30 |
temp = ""
|
| 31 |
input_prompt = f"[INST] <<SYS>>\n{system_message}\n<</SYS>>\n\n "
|
| 32 |
for interaction in history:
|
| 33 |
+
input_prompt = (
|
| 34 |
+
input_prompt
|
| 35 |
+
+ str(interaction[0])
|
| 36 |
+
+ " [/INST] "
|
| 37 |
+
+ str(interaction[1])
|
| 38 |
+
+ " </s><s> [INST] "
|
| 39 |
+
)
|
| 40 |
|
| 41 |
input_prompt = input_prompt + str(message) + " [/INST] "
|
| 42 |
|
|
|
|
| 68 |
title="llama-cpp-python on GPU",
|
| 69 |
description="Running LLM with https://github.com/abetlen/llama-cpp-python",
|
| 70 |
examples=[
|
| 71 |
+
["How to setup a human base on Mars? Give short answer."],
|
| 72 |
+
["Explain theory of relativity to me like I’m 8 years old."],
|
| 73 |
+
["What is 9,000 * 9,000?"],
|
| 74 |
+
["Write a pun-filled happy birthday message to my friend Alex."],
|
| 75 |
+
["Justify why a penguin might make a good king of the jungle."],
|
| 76 |
],
|
| 77 |
cache_examples=False,
|
| 78 |
retry_btn=None,
|