Spaces:
Running
on
Zero
Running
on
Zero
Update app.py
Browse files
app.py
CHANGED
@@ -117,6 +117,12 @@ moondream = AutoModelForCausalLM.from_pretrained(
|
|
117 |
tokenizer_md = AutoTokenizer.from_pretrained(MODEL_ID_MD, revision=REVISION_MD)
|
118 |
|
119 |
|
|
|
|
|
|
|
|
|
|
|
|
|
120 |
# --- PDF Generation and Preview Utility Function ---
|
121 |
def generate_and_preview_pdf(image: Image.Image, text_content: str, font_size: int, line_spacing: float, alignment: str, image_size: str):
|
122 |
"""
|
@@ -225,6 +231,7 @@ def process_document_stream(
|
|
225 |
elif model_name == "VLAA-Thinker-Qwen2VL-2B(reason)": processor, model = processor_i, model_i
|
226 |
elif model_name == "Nanonets-OCR-s(ocr)": processor, model = processor_a, model_a
|
227 |
elif model_name == "Megalodon-OCR-Sync-0713(ocr)": processor, model = processor_x, model_x
|
|
|
228 |
else:
|
229 |
yield "Invalid model selected.", ""
|
230 |
return
|
@@ -282,7 +289,7 @@ def create_gradio_interface():
|
|
282 |
with gr.Column(scale=1):
|
283 |
model_choice = gr.Dropdown(
|
284 |
choices=["LFM2-VL-450M(fast)", "LFM2-VL-1.6B(fast)", "SmolVLM-Instruct-250M(smol)", "Moondream2(vision)", "ShotVL-3B(cinematic)", "Megalodon-OCR-Sync-0713(ocr)",
|
285 |
-
"VLAA-Thinker-Qwen2VL-2B(reason)", "MonkeyOCR-pro-1.2B(ocr)", "Nanonets-OCR-s(ocr)"],
|
286 |
label="Select Model", value= "LFM2-VL-450M(fast)"
|
287 |
)
|
288 |
prompt_input = gr.Textbox(label="Query Input", placeholder="✦︎ Enter your query", value="Describe the image!")
|
|
|
117 |
tokenizer_md = AutoTokenizer.from_pretrained(MODEL_ID_MD, revision=REVISION_MD)
|
118 |
|
119 |
|
120 |
+
MODEL_ID_N = "MrLight/dse-qwen2-2b-mrl-v1"
|
121 |
+
processor_n = AutoProcessor.from_pretrained(MODEL_ID_N, trust_remote_code=True)
|
122 |
+
model_n = Qwen2VLForConditionalGeneration.from_pretrained(
|
123 |
+
MODEL_ID_N, trust_remote_code=True, torch_dtype=torch.float16
|
124 |
+
).to(device).eval()
|
125 |
+
|
126 |
# --- PDF Generation and Preview Utility Function ---
|
127 |
def generate_and_preview_pdf(image: Image.Image, text_content: str, font_size: int, line_spacing: float, alignment: str, image_size: str):
|
128 |
"""
|
|
|
231 |
elif model_name == "VLAA-Thinker-Qwen2VL-2B(reason)": processor, model = processor_i, model_i
|
232 |
elif model_name == "Nanonets-OCR-s(ocr)": processor, model = processor_a, model_a
|
233 |
elif model_name == "Megalodon-OCR-Sync-0713(ocr)": processor, model = processor_x, model_x
|
234 |
+
elif model_name == "dse-qwen2-2b-mrl-v1(dse)": processor, model = processor_n, model_n
|
235 |
else:
|
236 |
yield "Invalid model selected.", ""
|
237 |
return
|
|
|
289 |
with gr.Column(scale=1):
|
290 |
model_choice = gr.Dropdown(
|
291 |
choices=["LFM2-VL-450M(fast)", "LFM2-VL-1.6B(fast)", "SmolVLM-Instruct-250M(smol)", "Moondream2(vision)", "ShotVL-3B(cinematic)", "Megalodon-OCR-Sync-0713(ocr)",
|
292 |
+
"VLAA-Thinker-Qwen2VL-2B(reason)", "MonkeyOCR-pro-1.2B(ocr)", "dse-qwen2-2b-mrl-v1(dse)", "Nanonets-OCR-s(ocr)"],
|
293 |
label="Select Model", value= "LFM2-VL-450M(fast)"
|
294 |
)
|
295 |
prompt_input = gr.Textbox(label="Query Input", placeholder="✦︎ Enter your query", value="Describe the image!")
|