Spaces:
Running
on
Zero
Running
on
Zero
Update app.py
Browse files
app.py
CHANGED
@@ -34,7 +34,7 @@ MAX_INPUT_TOKEN_LENGTH = int(os.getenv("MAX_INPUT_TOKEN_LENGTH", "4096"))
|
|
34 |
|
35 |
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
|
36 |
|
37 |
-
#
|
38 |
model_id = "prithivMLmods/FastThink-0.5B-Tiny"
|
39 |
tokenizer = AutoTokenizer.from_pretrained(model_id)
|
40 |
model = AutoModelForCausalLM.from_pretrained(
|
@@ -53,7 +53,7 @@ TTS_VOICES = [
|
|
53 |
"en-US-JasonNeural", # @tts6
|
54 |
]
|
55 |
|
56 |
-
#
|
57 |
MODEL_ID = "prithivMLmods/Qwen2-VL-OCR-2B-Instruct"
|
58 |
processor = AutoProcessor.from_pretrained(MODEL_ID, trust_remote_code=True)
|
59 |
model_m = Qwen2VLForConditionalGeneration.from_pretrained(
|
@@ -70,12 +70,11 @@ async def text_to_speech(text: str, voice: str, output_file="output.mp3"):
|
|
70 |
|
71 |
def clean_chat_history(chat_history):
|
72 |
"""
|
73 |
-
Filter out any entries whose content is not a string.
|
74 |
-
This
|
75 |
"""
|
76 |
cleaned = []
|
77 |
for msg in chat_history:
|
78 |
-
# Only keep dict messages that have a string 'content'
|
79 |
if isinstance(msg, dict) and isinstance(msg.get("content"), str):
|
80 |
cleaned.append(msg)
|
81 |
return cleaned
|
@@ -91,14 +90,13 @@ def generate(
|
|
91 |
repetition_penalty: float = 1.2,
|
92 |
):
|
93 |
"""
|
94 |
-
Generates
|
95 |
-
If the
|
96 |
-
(clearing any non-text outputs). Otherwise, the chat history is cleaned to include only text.
|
97 |
"""
|
98 |
text = input_dict["text"]
|
99 |
files = input_dict.get("files", [])
|
100 |
|
101 |
-
#
|
102 |
if len(files) > 1:
|
103 |
images = [load_image(image) for image in files]
|
104 |
elif len(files) == 1:
|
@@ -106,25 +104,23 @@ def generate(
|
|
106 |
else:
|
107 |
images = []
|
108 |
|
109 |
-
# Check for TTS prefix
|
110 |
tts_prefix = "@tts"
|
111 |
is_tts = any(text.strip().lower().startswith(f"{tts_prefix}{i}") for i in range(1, 7))
|
112 |
voice_index = next((i for i in range(1, 7) if text.strip().lower().startswith(f"{tts_prefix}{i}")), None)
|
113 |
-
|
114 |
if is_tts and voice_index:
|
115 |
voice = TTS_VOICES[voice_index - 1]
|
116 |
text = text.replace(f"{tts_prefix}{voice_index}", "").strip()
|
117 |
-
# Clear any previous chat history
|
118 |
conversation = [{"role": "user", "content": text}]
|
119 |
else:
|
120 |
voice = None
|
121 |
text = text.replace(tts_prefix, "").strip()
|
122 |
-
# Clean the chat history to include only messages with string content
|
123 |
conversation = clean_chat_history(chat_history)
|
124 |
conversation.append({"role": "user", "content": text})
|
125 |
|
126 |
-
# Multimodal branch if images are provided
|
127 |
if images:
|
|
|
128 |
messages = [{
|
129 |
"role": "user",
|
130 |
"content": [
|
@@ -134,9 +130,8 @@ def generate(
|
|
134 |
}]
|
135 |
prompt = processor.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
|
136 |
inputs = processor(text=[prompt], images=images, return_tensors="pt", padding=True).to("cuda")
|
137 |
-
|
138 |
streamer = TextIteratorStreamer(processor, skip_prompt=True, skip_special_tokens=True)
|
139 |
-
generation_kwargs =
|
140 |
thread = Thread(target=model_m.generate, kwargs=generation_kwargs)
|
141 |
thread.start()
|
142 |
|
@@ -154,19 +149,18 @@ def generate(
|
|
154 |
input_ids = input_ids[:, -MAX_INPUT_TOKEN_LENGTH:]
|
155 |
gr.Warning(f"Trimmed input from conversation as it was longer than {MAX_INPUT_TOKEN_LENGTH} tokens.")
|
156 |
input_ids = input_ids.to(model.device)
|
157 |
-
|
158 |
streamer = TextIteratorStreamer(tokenizer, timeout=20.0, skip_prompt=True, skip_special_tokens=True)
|
159 |
-
|
160 |
-
|
161 |
-
streamer
|
162 |
-
max_new_tokens
|
163 |
-
do_sample
|
164 |
-
top_p
|
165 |
-
top_k
|
166 |
-
temperature
|
167 |
-
num_beams
|
168 |
-
repetition_penalty
|
169 |
-
|
170 |
t = Thread(target=model.generate, kwargs=generation_kwargs)
|
171 |
t.start()
|
172 |
|
@@ -176,7 +170,6 @@ def generate(
|
|
176 |
yield "".join(outputs)
|
177 |
|
178 |
final_response = "".join(outputs)
|
179 |
-
# Yield text response first
|
180 |
yield final_response
|
181 |
|
182 |
if is_tts and voice:
|
|
|
34 |
|
35 |
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
|
36 |
|
37 |
+
# Load text-only model and tokenizer
|
38 |
model_id = "prithivMLmods/FastThink-0.5B-Tiny"
|
39 |
tokenizer = AutoTokenizer.from_pretrained(model_id)
|
40 |
model = AutoModelForCausalLM.from_pretrained(
|
|
|
53 |
"en-US-JasonNeural", # @tts6
|
54 |
]
|
55 |
|
56 |
+
# Load multimodal (OCR) model and processor
|
57 |
MODEL_ID = "prithivMLmods/Qwen2-VL-OCR-2B-Instruct"
|
58 |
processor = AutoProcessor.from_pretrained(MODEL_ID, trust_remote_code=True)
|
59 |
model_m = Qwen2VLForConditionalGeneration.from_pretrained(
|
|
|
70 |
|
71 |
def clean_chat_history(chat_history):
|
72 |
"""
|
73 |
+
Filter out any chat entries whose "content" is not a string.
|
74 |
+
This helps prevent errors when concatenating previous messages.
|
75 |
"""
|
76 |
cleaned = []
|
77 |
for msg in chat_history:
|
|
|
78 |
if isinstance(msg, dict) and isinstance(msg.get("content"), str):
|
79 |
cleaned.append(msg)
|
80 |
return cleaned
|
|
|
90 |
repetition_penalty: float = 1.2,
|
91 |
):
|
92 |
"""
|
93 |
+
Generates chatbot responses with support for multimodal input and TTS.
|
94 |
+
If the query starts with an @tts command (e.g. "@tts1"), previous chat history is cleared.
|
|
|
95 |
"""
|
96 |
text = input_dict["text"]
|
97 |
files = input_dict.get("files", [])
|
98 |
|
99 |
+
# Process image files if provided
|
100 |
if len(files) > 1:
|
101 |
images = [load_image(image) for image in files]
|
102 |
elif len(files) == 1:
|
|
|
104 |
else:
|
105 |
images = []
|
106 |
|
|
|
107 |
tts_prefix = "@tts"
|
108 |
is_tts = any(text.strip().lower().startswith(f"{tts_prefix}{i}") for i in range(1, 7))
|
109 |
voice_index = next((i for i in range(1, 7) if text.strip().lower().startswith(f"{tts_prefix}{i}")), None)
|
110 |
+
|
111 |
if is_tts and voice_index:
|
112 |
voice = TTS_VOICES[voice_index - 1]
|
113 |
text = text.replace(f"{tts_prefix}{voice_index}", "").strip()
|
114 |
+
# Clear any previous chat history to avoid concatenation issues
|
115 |
conversation = [{"role": "user", "content": text}]
|
116 |
else:
|
117 |
voice = None
|
118 |
text = text.replace(tts_prefix, "").strip()
|
|
|
119 |
conversation = clean_chat_history(chat_history)
|
120 |
conversation.append({"role": "user", "content": text})
|
121 |
|
|
|
122 |
if images:
|
123 |
+
# Multimodal branch using the OCR model
|
124 |
messages = [{
|
125 |
"role": "user",
|
126 |
"content": [
|
|
|
130 |
}]
|
131 |
prompt = processor.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
|
132 |
inputs = processor(text=[prompt], images=images, return_tensors="pt", padding=True).to("cuda")
|
|
|
133 |
streamer = TextIteratorStreamer(processor, skip_prompt=True, skip_special_tokens=True)
|
134 |
+
generation_kwargs = {**inputs, "streamer": streamer, "max_new_tokens": max_new_tokens}
|
135 |
thread = Thread(target=model_m.generate, kwargs=generation_kwargs)
|
136 |
thread.start()
|
137 |
|
|
|
149 |
input_ids = input_ids[:, -MAX_INPUT_TOKEN_LENGTH:]
|
150 |
gr.Warning(f"Trimmed input from conversation as it was longer than {MAX_INPUT_TOKEN_LENGTH} tokens.")
|
151 |
input_ids = input_ids.to(model.device)
|
|
|
152 |
streamer = TextIteratorStreamer(tokenizer, timeout=20.0, skip_prompt=True, skip_special_tokens=True)
|
153 |
+
generation_kwargs = {
|
154 |
+
"input_ids": input_ids,
|
155 |
+
"streamer": streamer,
|
156 |
+
"max_new_tokens": max_new_tokens,
|
157 |
+
"do_sample": True,
|
158 |
+
"top_p": top_p,
|
159 |
+
"top_k": top_k,
|
160 |
+
"temperature": temperature,
|
161 |
+
"num_beams": 1,
|
162 |
+
"repetition_penalty": repetition_penalty,
|
163 |
+
}
|
164 |
t = Thread(target=model.generate, kwargs=generation_kwargs)
|
165 |
t.start()
|
166 |
|
|
|
170 |
yield "".join(outputs)
|
171 |
|
172 |
final_response = "".join(outputs)
|
|
|
173 |
yield final_response
|
174 |
|
175 |
if is_tts and voice:
|