Spaces:
Paused
Paused
Update app.py
Browse filesinitial commit
app.py
CHANGED
|
@@ -1,11 +1,11 @@
|
|
| 1 |
import gradio as gr
|
| 2 |
import os
|
| 3 |
-
import torch
|
| 4 |
-
import
|
| 5 |
from transformers import AutoModelForCausalLM
|
| 6 |
from huggingface_hub import login
|
| 7 |
-
import subprocess
|
| 8 |
|
|
|
|
| 9 |
subprocess.run(
|
| 10 |
"pip install flash-attn --no-build-isolation",
|
| 11 |
env={"FLASH_ATTENTION_SKIP_CUDA_BUILD": "TRUE"},
|
|
@@ -15,34 +15,38 @@ subprocess.run(
|
|
| 15 |
hf_token = os.getenv("HF_TOKEN")
|
| 16 |
login(token=hf_token, add_to_git_credential=True)
|
| 17 |
|
|
|
|
| 18 |
@spaces.GPU
|
| 19 |
def get_model_summary(model_name):
|
| 20 |
-
# Check if CUDA is available and set the device accordingly
|
| 21 |
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
| 22 |
-
|
| 23 |
-
# Load the model and move it to the selected device
|
| 24 |
model = AutoModelForCausalLM.from_pretrained(model_name, trust_remote_code=True).to(device)
|
| 25 |
-
|
| 26 |
-
# Return the model's architecture as a string
|
| 27 |
return str(model)
|
| 28 |
|
| 29 |
-
# Create the Gradio interface
|
| 30 |
-
|
| 31 |
-
|
| 32 |
-
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
|
| 36 |
-
|
| 37 |
-
|
| 38 |
-
|
| 39 |
-
|
| 40 |
-
|
| 41 |
-
|
| 42 |
-
|
| 43 |
-
|
| 44 |
-
|
| 45 |
-
|
| 46 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 47 |
|
| 48 |
-
interface
|
|
|
|
|
|
| 1 |
import gradio as gr
|
| 2 |
import os
|
| 3 |
+
import torch
|
| 4 |
+
import subprocess
|
| 5 |
from transformers import AutoModelForCausalLM
|
| 6 |
from huggingface_hub import login
|
|
|
|
| 7 |
|
| 8 |
+
# Install required package
|
| 9 |
subprocess.run(
|
| 10 |
"pip install flash-attn --no-build-isolation",
|
| 11 |
env={"FLASH_ATTENTION_SKIP_CUDA_BUILD": "TRUE"},
|
|
|
|
| 15 |
hf_token = os.getenv("HF_TOKEN")
|
| 16 |
login(token=hf_token, add_to_git_credential=True)
|
| 17 |
|
| 18 |
+
# Function to get the model summary
|
| 19 |
@spaces.GPU
|
| 20 |
def get_model_summary(model_name):
|
|
|
|
| 21 |
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
|
|
|
|
|
|
| 22 |
model = AutoModelForCausalLM.from_pretrained(model_name, trust_remote_code=True).to(device)
|
|
|
|
|
|
|
| 23 |
return str(model)
|
| 24 |
|
| 25 |
+
# Create the Gradio Blocks interface
|
| 26 |
+
with gr.Blocks() as demo:
|
| 27 |
+
with gr.Row():
|
| 28 |
+
with gr.Column():
|
| 29 |
+
textbox = gr.Textbox(label="Model Name")
|
| 30 |
+
examples = gr.Examples(
|
| 31 |
+
examples=[
|
| 32 |
+
["google/gemma-7b"],
|
| 33 |
+
["microsoft/Phi-3-mini-4k-instruct"],
|
| 34 |
+
["meta-llama/Meta-Llama-3-8B"],
|
| 35 |
+
["mistralai/Mistral-7B-Instruct-v0.3"],
|
| 36 |
+
["vikhyatk/moondream2"],
|
| 37 |
+
["microsoft/Phi-3-vision-128k-instruct"],
|
| 38 |
+
["openbmb/MiniCPM-Llama3-V-2_5"],
|
| 39 |
+
["google/paligemma-3b-mix-224"],
|
| 40 |
+
["HuggingFaceM4/idefics2-8b-chatty"],
|
| 41 |
+
["mistralai/Codestral-22B-v0.1"]
|
| 42 |
+
],
|
| 43 |
+
inputs=textbox
|
| 44 |
+
)
|
| 45 |
+
submit_button = gr.Button("Submit")
|
| 46 |
+
with gr.Column():
|
| 47 |
+
output = gr.Textbox(label="Output", lines=20)
|
| 48 |
+
|
| 49 |
+
submit_button.click(fn=get_model_summary, inputs=textbox, outputs=output)
|
| 50 |
|
| 51 |
+
# Launch the interface
|
| 52 |
+
demo.launch()
|