Upload Moondream
Browse files- config.json +1 -1
- model.safetensors +3 -0
- moondream.py +17 -10
- vision_encoder.py +14 -28
config.json
CHANGED
@@ -10,6 +10,6 @@
|
|
10 |
"phi_config": {
|
11 |
"model_type": "phi-msft"
|
12 |
},
|
13 |
-
"torch_dtype": "
|
14 |
"transformers_version": "4.36.2"
|
15 |
}
|
|
|
10 |
"phi_config": {
|
11 |
"model_type": "phi-msft"
|
12 |
},
|
13 |
+
"torch_dtype": "float16",
|
14 |
"transformers_version": "4.36.2"
|
15 |
}
|
model.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:892e51df302d98a83974761c4f386caddbad2edd0e84f228d9935b4aed33ee25
|
3 |
+
size 3715037856
|
moondream.py
CHANGED
@@ -1,10 +1,12 @@
|
|
1 |
import torch
|
|
|
2 |
from .vision_encoder import VisionEncoder
|
3 |
-
from .text_model import TextModel
|
4 |
from .configuration_moondream import MoondreamConfig
|
5 |
from transformers import PreTrainedModel
|
6 |
import re
|
7 |
|
|
|
|
|
8 |
|
9 |
class Moondream(PreTrainedModel):
|
10 |
config_class = MoondreamConfig
|
@@ -12,11 +14,16 @@ class Moondream(PreTrainedModel):
|
|
12 |
def __init__(self, config):
|
13 |
super().__init__(config)
|
14 |
self.vision_encoder = VisionEncoder()
|
15 |
-
|
|
|
|
|
|
|
|
|
|
|
16 |
|
17 |
@property
|
18 |
def device(self):
|
19 |
-
return self.text_model.
|
20 |
|
21 |
def encode_image(self, image):
|
22 |
return self.vision_encoder(image)
|
@@ -27,22 +34,22 @@ class Moondream(PreTrainedModel):
|
|
27 |
txt, return_tensors="pt", add_special_tokens=False
|
28 |
).input_ids.to(self.device)
|
29 |
|
|
|
|
|
30 |
# Add BOS token
|
31 |
embeds = []
|
32 |
embeds.append(
|
33 |
-
self.
|
34 |
-
(torch.tensor([[tokenizer.bos_token_id]], device=self.device))
|
35 |
-
)
|
36 |
)
|
37 |
|
38 |
if "<image>" not in prompt:
|
39 |
-
embeds.append(
|
40 |
else:
|
41 |
assert prompt.count("<image>") == 1
|
42 |
before, after = prompt.split("<image>")
|
43 |
-
embeds.append(
|
44 |
embeds.append(image_embeds.to(self.device))
|
45 |
-
embeds.append(
|
46 |
|
47 |
return torch.cat(embeds, dim=1)
|
48 |
|
@@ -67,7 +74,7 @@ class Moondream(PreTrainedModel):
|
|
67 |
|
68 |
with torch.no_grad():
|
69 |
inputs_embeds = self.input_embeds(prompt, image_embeds, tokenizer)
|
70 |
-
output_ids = self.text_model.
|
71 |
inputs_embeds=inputs_embeds, **generate_config
|
72 |
)
|
73 |
|
|
|
1 |
import torch
|
2 |
+
from torch import nn
|
3 |
from .vision_encoder import VisionEncoder
|
|
|
4 |
from .configuration_moondream import MoondreamConfig
|
5 |
from transformers import PreTrainedModel
|
6 |
import re
|
7 |
|
8 |
+
from .modeling_phi import PhiForCausalLM
|
9 |
+
from .configuration_moondream import PhiConfig
|
10 |
|
11 |
class Moondream(PreTrainedModel):
|
12 |
config_class = MoondreamConfig
|
|
|
14 |
def __init__(self, config):
|
15 |
super().__init__(config)
|
16 |
self.vision_encoder = VisionEncoder()
|
17 |
+
|
18 |
+
if type(config.phi_config) == dict:
|
19 |
+
phi_config = PhiConfig(**config.phi_config)
|
20 |
+
else:
|
21 |
+
phi_config = config.phi_config
|
22 |
+
self.text_model = PhiForCausalLM(phi_config)
|
23 |
|
24 |
@property
|
25 |
def device(self):
|
26 |
+
return self.text_model.device
|
27 |
|
28 |
def encode_image(self, image):
|
29 |
return self.vision_encoder(image)
|
|
|
34 |
txt, return_tensors="pt", add_special_tokens=False
|
35 |
).input_ids.to(self.device)
|
36 |
|
37 |
+
text_emb = self.text_model.get_input_embeddings()
|
38 |
+
|
39 |
# Add BOS token
|
40 |
embeds = []
|
41 |
embeds.append(
|
42 |
+
text_emb((torch.tensor([[tokenizer.bos_token_id]], device=self.device)))
|
|
|
|
|
43 |
)
|
44 |
|
45 |
if "<image>" not in prompt:
|
46 |
+
embeds.append(text_emb(_tokenize(prompt)))
|
47 |
else:
|
48 |
assert prompt.count("<image>") == 1
|
49 |
before, after = prompt.split("<image>")
|
50 |
+
embeds.append(text_emb(_tokenize(f"{before}<image>")))
|
51 |
embeds.append(image_embeds.to(self.device))
|
52 |
+
embeds.append(text_emb(_tokenize(f"</image>{after}")))
|
53 |
|
54 |
return torch.cat(embeds, dim=1)
|
55 |
|
|
|
74 |
|
75 |
with torch.no_grad():
|
76 |
inputs_embeds = self.input_embeds(prompt, image_embeds, tokenizer)
|
77 |
+
output_ids = self.text_model.generate(
|
78 |
inputs_embeds=inputs_embeds, **generate_config
|
79 |
)
|
80 |
|
vision_encoder.py
CHANGED
@@ -80,23 +80,18 @@ class VisionProjection(nn.Module):
|
|
80 |
model_dim = 2048
|
81 |
hidden_dim = model_dim * 4
|
82 |
|
83 |
-
self.
|
84 |
-
self.mlp2 = MLP(model_dim, hidden_dim, model_dim)
|
85 |
-
self.ln = nn.LayerNorm(model_dim)
|
86 |
|
87 |
@property
|
88 |
def device(self):
|
89 |
-
return self.
|
90 |
|
91 |
def forward(self, x):
|
92 |
-
|
93 |
-
x = self.ln(x)
|
94 |
-
x = x + self.mlp2(x)
|
95 |
-
return x
|
96 |
|
97 |
|
98 |
-
class
|
99 |
-
def __init__(self):
|
100 |
super().__init__()
|
101 |
|
102 |
self.encoder = ModelHolder(
|
@@ -109,17 +104,6 @@ class VisionTower(nn.Module):
|
|
109 |
|
110 |
self.projection = VisionProjection()
|
111 |
|
112 |
-
def forward(self, x):
|
113 |
-
x = self.encoder(x)
|
114 |
-
x = self.projection(x)
|
115 |
-
return x
|
116 |
-
|
117 |
-
|
118 |
-
class VisionEncoder(nn.Module):
|
119 |
-
def __init__(self) -> None:
|
120 |
-
super().__init__()
|
121 |
-
|
122 |
-
self.model = VisionTower()
|
123 |
self.preprocess = Compose(
|
124 |
[
|
125 |
Resize(size=(378, 378), interpolation=InterpolationMode.BICUBIC),
|
@@ -131,20 +115,22 @@ class VisionEncoder(nn.Module):
|
|
131 |
|
132 |
@property
|
133 |
def device(self):
|
134 |
-
return self.
|
135 |
|
136 |
@property
|
137 |
def dtype(self):
|
138 |
-
return self.
|
139 |
|
140 |
def __call__(self, image: Image) -> torch.Tensor:
|
141 |
with torch.no_grad():
|
142 |
-
|
143 |
self.preprocess(image.convert("RGB"))
|
144 |
.unsqueeze(0)
|
145 |
.to(self.device, dtype=self.dtype)
|
146 |
)
|
147 |
-
|
148 |
-
|
149 |
-
)
|
150 |
-
|
|
|
|
|
|
80 |
model_dim = 2048
|
81 |
hidden_dim = model_dim * 4
|
82 |
|
83 |
+
self.mlp = MLP(image_embedding_dim, hidden_dim, model_dim)
|
|
|
|
|
84 |
|
85 |
@property
|
86 |
def device(self):
|
87 |
+
return self.mlp.fc1.weight.device
|
88 |
|
89 |
def forward(self, x):
|
90 |
+
return self.mlp(x)
|
|
|
|
|
|
|
91 |
|
92 |
|
93 |
+
class VisionEncoder(nn.Module):
|
94 |
+
def __init__(self) -> None:
|
95 |
super().__init__()
|
96 |
|
97 |
self.encoder = ModelHolder(
|
|
|
104 |
|
105 |
self.projection = VisionProjection()
|
106 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
107 |
self.preprocess = Compose(
|
108 |
[
|
109 |
Resize(size=(378, 378), interpolation=InterpolationMode.BICUBIC),
|
|
|
115 |
|
116 |
@property
|
117 |
def device(self):
|
118 |
+
return self.projection.mlp.fc1.weight.device
|
119 |
|
120 |
@property
|
121 |
def dtype(self):
|
122 |
+
return self.projection.mlp.fc1.weight.dtype
|
123 |
|
124 |
def __call__(self, image: Image) -> torch.Tensor:
|
125 |
with torch.no_grad():
|
126 |
+
x = (
|
127 |
self.preprocess(image.convert("RGB"))
|
128 |
.unsqueeze(0)
|
129 |
.to(self.device, dtype=self.dtype)
|
130 |
)
|
131 |
+
x = rearrange(x, "b c (h p1) (w p2) -> b (h w) (c p1 p2)", p1=14, p2=14)
|
132 |
+
|
133 |
+
x = self.encoder(x)
|
134 |
+
x = self.projection(x)
|
135 |
+
|
136 |
+
return x
|