jacob-valdez
commited on
Commit
•
68a6c21
1
Parent(s):
c539103
Update README.md
Browse files
README.md
CHANGED
@@ -11,4 +11,64 @@ license: "apache-2.0"
|
|
11 |
---
|
12 |
# Model Card
|
13 |
|
14 |
-
`blenderbot-small-tflite` is a tflite version of `blenderbot-small-90M` I converted for my UTA CSE3310 class. See the repo at [https://github.com/kmosoti/DesparadosAEYE](https://github.com/kmosoti/DesparadosAEYE) and the conversion process [here](https://colab.research.google.com/drive/1YRLxI92MdpUV0d-W5ToQfe3_J_9a4PF4?usp=sharing).
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
11 |
---
|
12 |
# Model Card
|
13 |
|
14 |
+
`blenderbot-small-tflite` is a tflite version of `blenderbot-small-90M` I converted for my UTA CSE3310 class. See the repo at [https://github.com/kmosoti/DesparadosAEYE](https://github.com/kmosoti/DesparadosAEYE) and the conversion process [here](https://colab.research.google.com/drive/1YRLxI92MdpUV0d-W5ToQfe3_J_9a4PF4?usp=sharing).
|
15 |
+
|
16 |
+
You have to right pad your user and model input integers to make them [32,]-shaped. Then indicate te true length with the 3rd and 4th params.
|
17 |
+
|
18 |
+
```python
|
19 |
+
display(interpreter.get_input_details())
|
20 |
+
display(interpreter.get_output_details())
|
21 |
+
```
|
22 |
+
|
23 |
+
```json
|
24 |
+
[{'dtype': numpy.int32,
|
25 |
+
'index': 0,
|
26 |
+
'name': 'input_tokens',
|
27 |
+
'quantization': (0.0, 0),
|
28 |
+
'quantization_parameters': {'quantized_dimension': 0,
|
29 |
+
'scales': array([], dtype=float32),
|
30 |
+
'zero_points': array([], dtype=int32)},
|
31 |
+
'shape': array([32], dtype=int32),
|
32 |
+
'shape_signature': array([32], dtype=int32),
|
33 |
+
'sparsity_parameters': {}},
|
34 |
+
{'dtype': numpy.int32,
|
35 |
+
'index': 1,
|
36 |
+
'name': 'decoder_input_tokens',
|
37 |
+
'quantization': (0.0, 0),
|
38 |
+
'quantization_parameters': {'quantized_dimension': 0,
|
39 |
+
'scales': array([], dtype=float32),
|
40 |
+
'zero_points': array([], dtype=int32)},
|
41 |
+
'shape': array([32], dtype=int32),
|
42 |
+
'shape_signature': array([32], dtype=int32),
|
43 |
+
'sparsity_parameters': {}},
|
44 |
+
{'dtype': numpy.int32,
|
45 |
+
'index': 2,
|
46 |
+
'name': 'input_len',
|
47 |
+
'quantization': (0.0, 0),
|
48 |
+
'quantization_parameters': {'quantized_dimension': 0,
|
49 |
+
'scales': array([], dtype=float32),
|
50 |
+
'zero_points': array([], dtype=int32)},
|
51 |
+
'shape': array([], dtype=int32),
|
52 |
+
'shape_signature': array([], dtype=int32),
|
53 |
+
'sparsity_parameters': {}},
|
54 |
+
{'dtype': numpy.int32,
|
55 |
+
'index': 3,
|
56 |
+
'name': 'decoder_input_len',
|
57 |
+
'quantization': (0.0, 0),
|
58 |
+
'quantization_parameters': {'quantized_dimension': 0,
|
59 |
+
'scales': array([], dtype=float32),
|
60 |
+
'zero_points': array([], dtype=int32)},
|
61 |
+
'shape': array([], dtype=int32),
|
62 |
+
'shape_signature': array([], dtype=int32),
|
63 |
+
'sparsity_parameters': {}}]
|
64 |
+
[{'dtype': numpy.int32,
|
65 |
+
'index': 3112,
|
66 |
+
'name': 'Identity',
|
67 |
+
'quantization': (0.0, 0),
|
68 |
+
'quantization_parameters': {'quantized_dimension': 0,
|
69 |
+
'scales': array([], dtype=float32),
|
70 |
+
'zero_points': array([], dtype=int32)},
|
71 |
+
'shape': array([], dtype=int32),
|
72 |
+
'shape_signature': array([], dtype=int32),
|
73 |
+
'sparsity_parameters': {}}]
|
74 |
+
```
|