Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -12,10 +12,10 @@ TITLE = """
|
|
12 |
<p><em>Once a pod has started, you can safely close this tab. The results are then privately uploaded to <a href="https://gist.github.com/">GitHub Gist</a>, and the pod is automatically destroyed.</em></p>
|
13 |
"""
|
14 |
|
15 |
-
def autoeval(BENCHMARK,
|
16 |
runpod.api_key = RUNPOD_TOKEN
|
17 |
pod = runpod.create_pod(
|
18 |
-
name=f"Eval {
|
19 |
image_name="runpod/pytorch:2.0.1-py3.10-cuda11.8.0-devel-ubuntu22.04",
|
20 |
gpu_type_id=GPU,
|
21 |
cloud_type=CLOUD_TYPE,
|
@@ -25,7 +25,7 @@ def autoeval(BENCHMARK, MODEL, GPU, NUMBER_OF_GPUS, CONTAINER_DISK, CLOUD_TYPE,
|
|
25 |
template_id="au6nz6emhk",
|
26 |
env={
|
27 |
"BENCHMARK": BENCHMARK,
|
28 |
-
"MODEL_ID":
|
29 |
"REPO": REPO,
|
30 |
"TRUST_REMOTE_CODE": TRUST_REMOTE_CODE,
|
31 |
"DEBUG": DEBUG,
|
@@ -42,7 +42,7 @@ with gr.Blocks() as demo:
|
|
42 |
gr.Textbox("", label="Model", value="mlabonne/NeuralBeagle14-7B", info="ID of the model you want to evaluate", placeholder="mlabonne/NeuralBeagle14-7B"),
|
43 |
gr.Dropdown(GPU_LIST, label="GPU", value="NVIDIA GeForce RTX 3090", info="Select your GPU to run the evaluation"),
|
44 |
gr.Slider(minimum=1, maximum=8, value=1, step=1, label="Number of GPUs", info="Number of GPUs to use"),
|
45 |
-
gr.Slider(minimum=50, maximum=500, value=
|
46 |
gr.Dropdown(["COMMUNITY", "SECURE"], value="COMMUNITY", label="Cloud type", info="Select your cloud type"),
|
47 |
gr.Textbox("https://github.com/mlabonne/llm-autoeval.git", label="LLM AutoEval repo", info="Link to your LLM AutoEval repo"),
|
48 |
gr.Checkbox(label="Trust remote code", value=False, info="Required for some models like phi-2"),
|
|
|
12 |
<p><em>Once a pod has started, you can safely close this tab. The results are then privately uploaded to <a href="https://gist.github.com/">GitHub Gist</a>, and the pod is automatically destroyed.</em></p>
|
13 |
"""
|
14 |
|
15 |
+
def autoeval(BENCHMARK, MODEL_ID, GPU, NUMBER_OF_GPUS, CONTAINER_DISK, CLOUD_TYPE, REPO, TRUST_REMOTE_CODE, DEBUG, GITHUB_API_TOKEN, RUNPOD_TOKEN):
|
16 |
runpod.api_key = RUNPOD_TOKEN
|
17 |
pod = runpod.create_pod(
|
18 |
+
name=f"Eval {MODEL_ID.split('/')[-1]} on {BENCHMARK.capitalize()}",
|
19 |
image_name="runpod/pytorch:2.0.1-py3.10-cuda11.8.0-devel-ubuntu22.04",
|
20 |
gpu_type_id=GPU,
|
21 |
cloud_type=CLOUD_TYPE,
|
|
|
25 |
template_id="au6nz6emhk",
|
26 |
env={
|
27 |
"BENCHMARK": BENCHMARK,
|
28 |
+
"MODEL_ID": MODEL_ID,
|
29 |
"REPO": REPO,
|
30 |
"TRUST_REMOTE_CODE": TRUST_REMOTE_CODE,
|
31 |
"DEBUG": DEBUG,
|
|
|
42 |
gr.Textbox("", label="Model", value="mlabonne/NeuralBeagle14-7B", info="ID of the model you want to evaluate", placeholder="mlabonne/NeuralBeagle14-7B"),
|
43 |
gr.Dropdown(GPU_LIST, label="GPU", value="NVIDIA GeForce RTX 3090", info="Select your GPU to run the evaluation"),
|
44 |
gr.Slider(minimum=1, maximum=8, value=1, step=1, label="Number of GPUs", info="Number of GPUs to use"),
|
45 |
+
gr.Slider(minimum=50, maximum=500, value=75, step=25, label="Container disk", info="Size of the container disk in GB"),
|
46 |
gr.Dropdown(["COMMUNITY", "SECURE"], value="COMMUNITY", label="Cloud type", info="Select your cloud type"),
|
47 |
gr.Textbox("https://github.com/mlabonne/llm-autoeval.git", label="LLM AutoEval repo", info="Link to your LLM AutoEval repo"),
|
48 |
gr.Checkbox(label="Trust remote code", value=False, info="Required for some models like phi-2"),
|