Spaces:
Running
on
Zero
Running
on
Zero
Update app.py
Browse files
app.py
CHANGED
@@ -8,8 +8,8 @@ import torch
|
|
8 |
from transformers import AutoTokenizer, AutoModelForCausalLM, TextIteratorStreamer
|
9 |
|
10 |
|
11 |
-
model_id = "
|
12 |
-
assistant_id = "
|
13 |
|
14 |
model = AutoModelForCausalLM.from_pretrained(model_id, load_in_4bit=True, device_map="auto")
|
15 |
assistant_model = AutoModelForCausalLM.from_pretrained(assistant_id).to(device=model.device, dtype=torch.float16)
|
@@ -61,9 +61,9 @@ def reset_textbox():
|
|
61 |
with gr.Blocks() as demo:
|
62 |
gr.Markdown(
|
63 |
"# 🤗 Assisted Generation Demo\n"
|
64 |
-
f"- Model: {model_id}
|
65 |
-
f"- Assistant Model: {assistant_id}
|
66 |
-
"- Recipe for speedup: a) >10x model size difference in parameters; b) assistant trained similarly; c) CPU is not a bottleneck"
|
67 |
)
|
68 |
|
69 |
with gr.Row():
|
|
|
8 |
from transformers import AutoTokenizer, AutoModelForCausalLM, TextIteratorStreamer
|
9 |
|
10 |
|
11 |
+
model_id = "meta-llama/Llama-3.1-8B"
|
12 |
+
assistant_id = "meta-llama/Llama-3.2-1B"
|
13 |
|
14 |
model = AutoModelForCausalLM.from_pretrained(model_id, load_in_4bit=True, device_map="auto")
|
15 |
assistant_model = AutoModelForCausalLM.from_pretrained(assistant_id).to(device=model.device, dtype=torch.float16)
|
|
|
61 |
with gr.Blocks() as demo:
|
62 |
gr.Markdown(
|
63 |
"# 🤗 Assisted Generation Demo\n"
|
64 |
+
f"- Model: {model_id}\n"
|
65 |
+
f"- Assistant Model: {assistant_id}\n"
|
66 |
+
"- Recipe for good speedup: a) >10x model size difference in parameters; b) assistant trained similarly; c) CPU is not a bottleneck"
|
67 |
)
|
68 |
|
69 |
with gr.Row():
|