Spaces:
Runtime error
Runtime error
ryanzhangfan
commited on
Commit
•
6380db8
1
Parent(s):
116a3d4
Update app.py
Browse files
app.py
CHANGED
@@ -24,6 +24,8 @@ subprocess.run(
|
|
24 |
shell=True,
|
25 |
)
|
26 |
|
|
|
|
|
27 |
# Model paths
|
28 |
EMU_GEN_HUB = "BAAI/Emu3-Gen"
|
29 |
EMU_CHAT_HUB = "BAAI/Emu3-Chat"
|
@@ -33,28 +35,28 @@ VQ_HUB = "BAAI/Emu3-VisionTokenizer"
|
|
33 |
# Emu3-Gen model and processor
|
34 |
gen_model = AutoModelForCausalLM.from_pretrained(
|
35 |
EMU_GEN_HUB,
|
36 |
-
device_map="
|
37 |
torch_dtype=torch.bfloat16,
|
38 |
attn_implementation="flash_attention_2",
|
39 |
trust_remote_code=True,
|
40 |
-
)
|
41 |
|
42 |
# Emu3-Chat model and processor
|
43 |
chat_model = AutoModelForCausalLM.from_pretrained(
|
44 |
EMU_CHAT_HUB,
|
45 |
-
device_map="
|
46 |
torch_dtype=torch.bfloat16,
|
47 |
attn_implementation="flash_attention_2",
|
48 |
trust_remote_code=True,
|
49 |
-
)
|
50 |
|
51 |
tokenizer = AutoTokenizer.from_pretrained(EMU_CHAT_HUB, trust_remote_code=True)
|
52 |
image_processor = AutoImageProcessor.from_pretrained(
|
53 |
VQ_HUB, trust_remote_code=True
|
54 |
)
|
55 |
image_tokenizer = AutoModel.from_pretrained(
|
56 |
-
VQ_HUB, device_map="
|
57 |
-
).eval()
|
58 |
processor = Emu3Processor(
|
59 |
image_processor, image_tokenizer, tokenizer
|
60 |
)
|
@@ -97,7 +99,7 @@ def generate_image(prompt):
|
|
97 |
UnbatchedClassifierFreeGuidanceLogitsProcessor(
|
98 |
classifier_free_guidance,
|
99 |
gen_model,
|
100 |
-
unconditional_ids=neg_inputs.input_ids.to(
|
101 |
),
|
102 |
PrefixConstrainedLogitsProcessor(
|
103 |
constrained_fn,
|
@@ -108,7 +110,7 @@ def generate_image(prompt):
|
|
108 |
|
109 |
# Generate
|
110 |
outputs = gen_model.generate(
|
111 |
-
pos_inputs.input_ids.to(
|
112 |
generation_config=GENERATION_CONFIG,
|
113 |
logits_processor=logits_processor,
|
114 |
)
|
@@ -139,7 +141,7 @@ def vision_language_understanding(image, text):
|
|
139 |
|
140 |
# Generate
|
141 |
outputs = chat_model.generate(
|
142 |
-
inputs.input_ids.to(
|
143 |
generation_config=GENERATION_CONFIG,
|
144 |
max_new_tokens=320,
|
145 |
)
|
|
|
24 |
shell=True,
|
25 |
)
|
26 |
|
27 |
+
device = "cuda" if torch.cuda.is_available() else "cpu"
|
28 |
+
|
29 |
# Model paths
|
30 |
EMU_GEN_HUB = "BAAI/Emu3-Gen"
|
31 |
EMU_CHAT_HUB = "BAAI/Emu3-Chat"
|
|
|
35 |
# Emu3-Gen model and processor
|
36 |
gen_model = AutoModelForCausalLM.from_pretrained(
|
37 |
EMU_GEN_HUB,
|
38 |
+
device_map="cpu",
|
39 |
torch_dtype=torch.bfloat16,
|
40 |
attn_implementation="flash_attention_2",
|
41 |
trust_remote_code=True,
|
42 |
+
).to(device)
|
43 |
|
44 |
# Emu3-Chat model and processor
|
45 |
chat_model = AutoModelForCausalLM.from_pretrained(
|
46 |
EMU_CHAT_HUB,
|
47 |
+
device_map="cpu",
|
48 |
torch_dtype=torch.bfloat16,
|
49 |
attn_implementation="flash_attention_2",
|
50 |
trust_remote_code=True,
|
51 |
+
).to(device)
|
52 |
|
53 |
tokenizer = AutoTokenizer.from_pretrained(EMU_CHAT_HUB, trust_remote_code=True)
|
54 |
image_processor = AutoImageProcessor.from_pretrained(
|
55 |
VQ_HUB, trust_remote_code=True
|
56 |
)
|
57 |
image_tokenizer = AutoModel.from_pretrained(
|
58 |
+
VQ_HUB, device_map="cpu", trust_remote_code=True
|
59 |
+
).eval().to(device)
|
60 |
processor = Emu3Processor(
|
61 |
image_processor, image_tokenizer, tokenizer
|
62 |
)
|
|
|
99 |
UnbatchedClassifierFreeGuidanceLogitsProcessor(
|
100 |
classifier_free_guidance,
|
101 |
gen_model,
|
102 |
+
unconditional_ids=neg_inputs.input_ids.to(device),
|
103 |
),
|
104 |
PrefixConstrainedLogitsProcessor(
|
105 |
constrained_fn,
|
|
|
110 |
|
111 |
# Generate
|
112 |
outputs = gen_model.generate(
|
113 |
+
pos_inputs.input_ids.to(device),
|
114 |
generation_config=GENERATION_CONFIG,
|
115 |
logits_processor=logits_processor,
|
116 |
)
|
|
|
141 |
|
142 |
# Generate
|
143 |
outputs = chat_model.generate(
|
144 |
+
inputs.input_ids.to(device),
|
145 |
generation_config=GENERATION_CONFIG,
|
146 |
max_new_tokens=320,
|
147 |
)
|