Spaces:
Build error
Build error
fix layout, max size back to 1, llama.cpp doesn't like parallel calls
Browse files- config.yml +1 -1
- tabbed.py +11 -23
config.yml
CHANGED
@@ -11,4 +11,4 @@ chat:
|
|
11 |
- "### User:"
|
12 |
queue:
|
13 |
max_size: 16
|
14 |
-
concurrency_count:
|
|
|
11 |
- "### User:"
|
12 |
queue:
|
13 |
max_size: 16
|
14 |
+
concurrency_count: 1
|
tabbed.py
CHANGED
@@ -70,26 +70,24 @@ def generate_text_instruct(input_text):
|
|
70 |
return output['choices'][0]['text']
|
71 |
|
72 |
|
73 |
-
instruct_description = f"""
|
74 |
-
### brought to you by OpenAccess AI Collective
|
75 |
-
- This is the [{config["repo"]}](https://huggingface.co/{config["repo"]}) model file [{config["file"]}](https://huggingface.co/{config["repo"]}/blob/main/{config["file"]})
|
76 |
-
- This Space uses GGML with GPU support, so it can quickly run larger models on smaller GPUs & VRAM.
|
77 |
-
- This is running on a smaller, shared GPU, so it may take a few seconds to respond.
|
78 |
-
- Due to a [missing feature in Gradio](https://github.com/gradio-app/gradio/issues/3914), the chatbot interface will not show you your status in the queue. If it's stuck, be patient.
|
79 |
-
- [Duplicate the Space](https://huggingface.co/spaces/openaccess-ai-collective/ggml-ui?duplicate=true) to skip the queue and run in a private space or to use your own GGML models.
|
80 |
-
- When using your own models, simply update the [config.yml](https://huggingface.co/spaces/openaccess-ai-collective/ggml-ui/blob/main/config.yml)
|
81 |
-
- Contribute at [https://github.com/OpenAccess-AI-Collective/ggml-webui](https://github.com/OpenAccess-AI-Collective/ggml-webui)
|
82 |
-
"""
|
83 |
-
|
84 |
instruct_interface = gr.Interface(
|
85 |
fn=generate_text_instruct,
|
86 |
inputs=gr.inputs.Textbox(lines= 10, label="Enter your input text"),
|
87 |
outputs=gr.outputs.Textbox(label="Output text"),
|
88 |
-
title="GGML UI Chatbot Demo",
|
89 |
-
description=instruct_description,
|
90 |
)
|
91 |
|
92 |
with gr.Blocks() as demo:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
93 |
with gr.Tab("Instruct"):
|
94 |
gr.Markdown("# GGML Spaces Instruct Demo")
|
95 |
instruct_interface.render()
|
@@ -108,16 +106,6 @@ with gr.Blocks() as demo:
|
|
108 |
clear = gr.Button(value="New topic", variant="secondary").style(full_width=False)
|
109 |
stop = gr.Button(value="Stop", variant="secondary").style(full_width=False)
|
110 |
with gr.Row():
|
111 |
-
with gr.Column():
|
112 |
-
gr.Markdown(f"""
|
113 |
-
### brought to you by OpenAccess AI Collective
|
114 |
-
- This is the [{config["repo"]}](https://huggingface.co/{config["repo"]}) model file [{config["file"]}](https://huggingface.co/{config["repo"]}/blob/main/{config["file"]})
|
115 |
-
- This Space uses GGML with GPU support, so it can quickly run larger models on smaller GPUs & VRAM.
|
116 |
-
- This is running on a smaller, shared GPU, so it may take a few seconds to respond.
|
117 |
-
- [Duplicate the Space](https://huggingface.co/spaces/openaccess-ai-collective/ggml-ui?duplicate=true) to skip the queue and run in a private space or to use your own GGML models.
|
118 |
-
- When using your own models, simply update the [config.yml](https://huggingface.co/spaces/openaccess-ai-collective/ggml-ui/blob/main/config.yml)
|
119 |
-
- Contribute at [https://github.com/OpenAccess-AI-Collective/ggml-webui](https://github.com/OpenAccess-AI-Collective/ggml-webui)
|
120 |
-
""")
|
121 |
with gr.Column():
|
122 |
max_tokens = gr.Slider(20, 1000, label="Max Tokens", step=20, value=300)
|
123 |
temperature = gr.Slider(0.2, 2.0, label="Temperature", step=0.1, value=0.2)
|
|
|
70 |
return output['choices'][0]['text']
|
71 |
|
72 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
73 |
instruct_interface = gr.Interface(
|
74 |
fn=generate_text_instruct,
|
75 |
inputs=gr.inputs.Textbox(lines= 10, label="Enter your input text"),
|
76 |
outputs=gr.outputs.Textbox(label="Output text"),
|
|
|
|
|
77 |
)
|
78 |
|
79 |
with gr.Blocks() as demo:
|
80 |
+
with gr.Row():
|
81 |
+
with gr.Column():
|
82 |
+
gr.Markdown(f"""
|
83 |
+
### brought to you by OpenAccess AI Collective
|
84 |
+
- This is the [{config["repo"]}](https://huggingface.co/{config["repo"]}) model file [{config["file"]}](https://huggingface.co/{config["repo"]}/blob/main/{config["file"]})
|
85 |
+
- This Space uses GGML with GPU support, so it can quickly run larger models on smaller GPUs & VRAM.
|
86 |
+
- This is running on a smaller, shared GPU, so it may take a few seconds to respond.
|
87 |
+
- [Duplicate the Space](https://huggingface.co/spaces/openaccess-ai-collective/ggml-ui?duplicate=true) to skip the queue and run in a private space or to use your own GGML models.
|
88 |
+
- When using your own models, simply update the [config.yml](https://huggingface.co/spaces/openaccess-ai-collective/ggml-ui/blob/main/config.yml)
|
89 |
+
- Contribute at [https://github.com/OpenAccess-AI-Collective/ggml-webui](https://github.com/OpenAccess-AI-Collective/ggml-webui)
|
90 |
+
""")
|
91 |
with gr.Tab("Instruct"):
|
92 |
gr.Markdown("# GGML Spaces Instruct Demo")
|
93 |
instruct_interface.render()
|
|
|
106 |
clear = gr.Button(value="New topic", variant="secondary").style(full_width=False)
|
107 |
stop = gr.Button(value="Stop", variant="secondary").style(full_width=False)
|
108 |
with gr.Row():
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
109 |
with gr.Column():
|
110 |
max_tokens = gr.Slider(20, 1000, label="Max Tokens", step=20, value=300)
|
111 |
temperature = gr.Slider(0.2, 2.0, label="Temperature", step=0.1, value=0.2)
|