Upload hybrid_test.ipynb
Browse files- hybrid_test.ipynb +410 -0
hybrid_test.ipynb
ADDED
@@ -0,0 +1,410 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"cells": [
|
3 |
+
{
|
4 |
+
"cell_type": "code",
|
5 |
+
"execution_count": 1,
|
6 |
+
"metadata": {
|
7 |
+
"id": "0paOn0yhDB63"
|
8 |
+
},
|
9 |
+
"outputs": [],
|
10 |
+
"source": [
|
11 |
+
"from hybrid_pipe import HybridQAPipeline\n",
|
12 |
+
"from transformers import pipeline\n",
|
13 |
+
"from transformers.pipelines import PIPELINE_REGISTRY\n",
|
14 |
+
"\n",
|
15 |
+
"from transformers import AutoModelForQuestionAnswering, TFAutoModelForQuestionAnswering\n"
|
16 |
+
]
|
17 |
+
},
|
18 |
+
{
|
19 |
+
"cell_type": "code",
|
20 |
+
"execution_count": 3,
|
21 |
+
"metadata": {
|
22 |
+
"id": "DuwOF8yjDB66"
|
23 |
+
},
|
24 |
+
"outputs": [],
|
25 |
+
"source": [
|
26 |
+
"# Register new pipe\n",
|
27 |
+
"PIPELINE_REGISTRY.register_pipeline(\n",
|
28 |
+
" \"hybrid-qa\",\n",
|
29 |
+
" pipeline_class=HybridQAPipeline,\n",
|
30 |
+
" pt_model=AutoModelForQuestionAnswering,\n",
|
31 |
+
" tf_model=TFAutoModelForQuestionAnswering\n",
|
32 |
+
")"
|
33 |
+
]
|
34 |
+
},
|
35 |
+
{
|
36 |
+
"cell_type": "code",
|
37 |
+
"execution_count": 4,
|
38 |
+
"metadata": {
|
39 |
+
"id": "pf_tBYQsDB67",
|
40 |
+
"colab": {
|
41 |
+
"base_uri": "https://localhost:8080/"
|
42 |
+
},
|
43 |
+
"outputId": "2d75ec1b-a844-441b-ca84-7859dd8eedc5"
|
44 |
+
},
|
45 |
+
"outputs": [
|
46 |
+
{
|
47 |
+
"output_type": "stream",
|
48 |
+
"name": "stderr",
|
49 |
+
"text": [
|
50 |
+
"You are using the default legacy behaviour of the <class 'transformers.models.t5.tokenization_t5.T5Tokenizer'>. This is expected, and simply means that the `legacy` (previous) behavior will be used so nothing changes for you. If you want to use the new behaviour, set `legacy=False`. This should only be set if you understand what it means, and thoroughly read the reason why this was added as explained in https://github.com/huggingface/transformers/pull/24565\n"
|
51 |
+
]
|
52 |
+
}
|
53 |
+
],
|
54 |
+
"source": [
|
55 |
+
"# Create pipe instance\n",
|
56 |
+
"# Note: the model specified here does not matter, we just need to\n",
|
57 |
+
"# pass something valid to satisfy the pipeline class=\n",
|
58 |
+
"hybrid_pipe = pipeline(\"hybrid-qa\", model='datarpit/distilbert-base-uncased-finetuned-natural-questions')"
|
59 |
+
]
|
60 |
+
},
|
61 |
+
{
|
62 |
+
"cell_type": "code",
|
63 |
+
"execution_count": 5,
|
64 |
+
"metadata": {
|
65 |
+
"colab": {
|
66 |
+
"base_uri": "https://localhost:8080/"
|
67 |
+
},
|
68 |
+
"id": "KKv6ZS2LDB67",
|
69 |
+
"outputId": "58f78991-1204-4714-af1c-bad70d120118"
|
70 |
+
},
|
71 |
+
"outputs": [
|
72 |
+
{
|
73 |
+
"output_type": "stream",
|
74 |
+
"name": "stderr",
|
75 |
+
"text": [
|
76 |
+
"Truncation was not explicitly activated but `max_length` is provided a specific value, please use `truncation=True` to explicitly truncate examples to max length. Defaulting to 'longest_first' truncation strategy. If you encode pairs of sequences (GLUE-style) with the tokenizer you can select this strategy more precisely by providing a specific strategy to `truncation`.\n",
|
77 |
+
"/usr/local/lib/python3.10/dist-packages/transformers/generation/utils.py:1141: UserWarning: Using the model-agnostic default `max_length` (=20) to control the generation length. We recommend setting `max_new_tokens` to control the maximum length of the generation.\n",
|
78 |
+
" warnings.warn(\n"
|
79 |
+
]
|
80 |
+
},
|
81 |
+
{
|
82 |
+
"output_type": "execute_result",
|
83 |
+
"data": {
|
84 |
+
"text/plain": [
|
85 |
+
"{'guess': 'Oslo', 'confidence': 2.0940363768613864e-14}"
|
86 |
+
]
|
87 |
+
},
|
88 |
+
"metadata": {},
|
89 |
+
"execution_count": 5
|
90 |
+
}
|
91 |
+
],
|
92 |
+
"source": [
|
93 |
+
"# Inference testing!\n",
|
94 |
+
"hybrid_pipe(question=\"What is the capital of Norway?\",context=\"The capital of Norway is Oslo\")"
|
95 |
+
]
|
96 |
+
},
|
97 |
+
{
|
98 |
+
"cell_type": "code",
|
99 |
+
"execution_count": 6,
|
100 |
+
"metadata": {
|
101 |
+
"colab": {
|
102 |
+
"base_uri": "https://localhost:8080/",
|
103 |
+
"height": 53
|
104 |
+
},
|
105 |
+
"id": "sgrDgs9-DB68",
|
106 |
+
"outputId": "7fe9f733-f19b-43cb-e68c-33e302b2be43"
|
107 |
+
},
|
108 |
+
"outputs": [
|
109 |
+
{
|
110 |
+
"output_type": "execute_result",
|
111 |
+
"data": {
|
112 |
+
"text/plain": [
|
113 |
+
"CommitInfo(commit_url='https://huggingface.co/justinhl/hybrid-qa/commit/7019d3e4971d6c754e9529b5a3de9a0425c3cccf', commit_message='Upload HybridQAPipeline', commit_description='', oid='7019d3e4971d6c754e9529b5a3de9a0425c3cccf', pr_url=None, pr_revision=None, pr_num=None)"
|
114 |
+
],
|
115 |
+
"application/vnd.google.colaboratory.intrinsic+json": {
|
116 |
+
"type": "string"
|
117 |
+
}
|
118 |
+
},
|
119 |
+
"metadata": {},
|
120 |
+
"execution_count": 6
|
121 |
+
}
|
122 |
+
],
|
123 |
+
"source": [
|
124 |
+
"# Pushing to hub\n",
|
125 |
+
"hybrid_pipe.push_to_hub(\"hybrid-qa\")"
|
126 |
+
]
|
127 |
+
},
|
128 |
+
{
|
129 |
+
"cell_type": "code",
|
130 |
+
"execution_count": 7,
|
131 |
+
"metadata": {
|
132 |
+
"colab": {
|
133 |
+
"base_uri": "https://localhost:8080/"
|
134 |
+
},
|
135 |
+
"id": "PPOf6vUhDB68",
|
136 |
+
"outputId": "1fa601e1-dfc7-4128-c430-44652916aa87"
|
137 |
+
},
|
138 |
+
"outputs": [
|
139 |
+
{
|
140 |
+
"output_type": "stream",
|
141 |
+
"name": "stderr",
|
142 |
+
"text": [
|
143 |
+
"Some weights of the model checkpoint at justinhl/hybrid-qa were not used when initializing DistilBertForQuestionAnswering: ['model_extractive.distilbert.embeddings.LayerNorm.bias', 'model_extractive.distilbert.embeddings.LayerNorm.weight', 'model_extractive.distilbert.embeddings.position_embeddings.weight', 'model_extractive.distilbert.embeddings.word_embeddings.weight', 'model_extractive.distilbert.transformer.layer.0.attention.k_lin.bias', 'model_extractive.distilbert.transformer.layer.0.attention.k_lin.weight', 'model_extractive.distilbert.transformer.layer.0.attention.out_lin.bias', 'model_extractive.distilbert.transformer.layer.0.attention.out_lin.weight', 'model_extractive.distilbert.transformer.layer.0.attention.q_lin.bias', 'model_extractive.distilbert.transformer.layer.0.attention.q_lin.weight', 'model_extractive.distilbert.transformer.layer.0.attention.v_lin.bias', 'model_extractive.distilbert.transformer.layer.0.attention.v_lin.weight', 'model_extractive.distilbert.transformer.layer.0.ffn.lin1.bias', 'model_extractive.distilbert.transformer.layer.0.ffn.lin1.weight', 'model_extractive.distilbert.transformer.layer.0.ffn.lin2.bias', 'model_extractive.distilbert.transformer.layer.0.ffn.lin2.weight', 'model_extractive.distilbert.transformer.layer.0.output_layer_norm.bias', 'model_extractive.distilbert.transformer.layer.0.output_layer_norm.weight', 'model_extractive.distilbert.transformer.layer.0.sa_layer_norm.bias', 'model_extractive.distilbert.transformer.layer.0.sa_layer_norm.weight', 'model_extractive.distilbert.transformer.layer.1.attention.k_lin.bias', 'model_extractive.distilbert.transformer.layer.1.attention.k_lin.weight', 'model_extractive.distilbert.transformer.layer.1.attention.out_lin.bias', 'model_extractive.distilbert.transformer.layer.1.attention.out_lin.weight', 'model_extractive.distilbert.transformer.layer.1.attention.q_lin.bias', 'model_extractive.distilbert.transformer.layer.1.attention.q_lin.weight', 'model_extractive.distilbert.transformer.layer.1.attention.v_lin.bias', 'model_extractive.distilbert.transformer.layer.1.attention.v_lin.weight', 'model_extractive.distilbert.transformer.layer.1.ffn.lin1.bias', 'model_extractive.distilbert.transformer.layer.1.ffn.lin1.weight', 'model_extractive.distilbert.transformer.layer.1.ffn.lin2.bias', 'model_extractive.distilbert.transformer.layer.1.ffn.lin2.weight', 'model_extractive.distilbert.transformer.layer.1.output_layer_norm.bias', 'model_extractive.distilbert.transformer.layer.1.output_layer_norm.weight', 'model_extractive.distilbert.transformer.layer.1.sa_layer_norm.bias', 'model_extractive.distilbert.transformer.layer.1.sa_layer_norm.weight', 'model_extractive.distilbert.transformer.layer.2.attention.k_lin.bias', 'model_extractive.distilbert.transformer.layer.2.attention.k_lin.weight', 'model_extractive.distilbert.transformer.layer.2.attention.out_lin.bias', 'model_extractive.distilbert.transformer.layer.2.attention.out_lin.weight', 'model_extractive.distilbert.transformer.layer.2.attention.q_lin.bias', 'model_extractive.distilbert.transformer.layer.2.attention.q_lin.weight', 'model_extractive.distilbert.transformer.layer.2.attention.v_lin.bias', 'model_extractive.distilbert.transformer.layer.2.attention.v_lin.weight', 'model_extractive.distilbert.transformer.layer.2.ffn.lin1.bias', 'model_extractive.distilbert.transformer.layer.2.ffn.lin1.weight', 'model_extractive.distilbert.transformer.layer.2.ffn.lin2.bias', 'model_extractive.distilbert.transformer.layer.2.ffn.lin2.weight', 'model_extractive.distilbert.transformer.layer.2.output_layer_norm.bias', 'model_extractive.distilbert.transformer.layer.2.output_layer_norm.weight', 'model_extractive.distilbert.transformer.layer.2.sa_layer_norm.bias', 'model_extractive.distilbert.transformer.layer.2.sa_layer_norm.weight', 'model_extractive.distilbert.transformer.layer.3.attention.k_lin.bias', 'model_extractive.distilbert.transformer.layer.3.attention.k_lin.weight', 'model_extractive.distilbert.transformer.layer.3.attention.out_lin.bias', 'model_extractive.distilbert.transformer.layer.3.attention.out_lin.weight', 'model_extractive.distilbert.transformer.layer.3.attention.q_lin.bias', 'model_extractive.distilbert.transformer.layer.3.attention.q_lin.weight', 'model_extractive.distilbert.transformer.layer.3.attention.v_lin.bias', 'model_extractive.distilbert.transformer.layer.3.attention.v_lin.weight', 'model_extractive.distilbert.transformer.layer.3.ffn.lin1.bias', 'model_extractive.distilbert.transformer.layer.3.ffn.lin1.weight', 'model_extractive.distilbert.transformer.layer.3.ffn.lin2.bias', 'model_extractive.distilbert.transformer.layer.3.ffn.lin2.weight', 'model_extractive.distilbert.transformer.layer.3.output_layer_norm.bias', 'model_extractive.distilbert.transformer.layer.3.output_layer_norm.weight', 'model_extractive.distilbert.transformer.layer.3.sa_layer_norm.bias', 'model_extractive.distilbert.transformer.layer.3.sa_layer_norm.weight', 'model_extractive.distilbert.transformer.layer.4.attention.k_lin.bias', 'model_extractive.distilbert.transformer.layer.4.attention.k_lin.weight', 'model_extractive.distilbert.transformer.layer.4.attention.out_lin.bias', 'model_extractive.distilbert.transformer.layer.4.attention.out_lin.weight', 'model_extractive.distilbert.transformer.layer.4.attention.q_lin.bias', 'model_extractive.distilbert.transformer.layer.4.attention.q_lin.weight', 'model_extractive.distilbert.transformer.layer.4.attention.v_lin.bias', 'model_extractive.distilbert.transformer.layer.4.attention.v_lin.weight', 'model_extractive.distilbert.transformer.layer.4.ffn.lin1.bias', 'model_extractive.distilbert.transformer.layer.4.ffn.lin1.weight', 'model_extractive.distilbert.transformer.layer.4.ffn.lin2.bias', 'model_extractive.distilbert.transformer.layer.4.ffn.lin2.weight', 'model_extractive.distilbert.transformer.layer.4.output_layer_norm.bias', 'model_extractive.distilbert.transformer.layer.4.output_layer_norm.weight', 'model_extractive.distilbert.transformer.layer.4.sa_layer_norm.bias', 'model_extractive.distilbert.transformer.layer.4.sa_layer_norm.weight', 'model_extractive.distilbert.transformer.layer.5.attention.k_lin.bias', 'model_extractive.distilbert.transformer.layer.5.attention.k_lin.weight', 'model_extractive.distilbert.transformer.layer.5.attention.out_lin.bias', 'model_extractive.distilbert.transformer.layer.5.attention.out_lin.weight', 'model_extractive.distilbert.transformer.layer.5.attention.q_lin.bias', 'model_extractive.distilbert.transformer.layer.5.attention.q_lin.weight', 'model_extractive.distilbert.transformer.layer.5.attention.v_lin.bias', 'model_extractive.distilbert.transformer.layer.5.attention.v_lin.weight', 'model_extractive.distilbert.transformer.layer.5.ffn.lin1.bias', 'model_extractive.distilbert.transformer.layer.5.ffn.lin1.weight', 'model_extractive.distilbert.transformer.layer.5.ffn.lin2.bias', 'model_extractive.distilbert.transformer.layer.5.ffn.lin2.weight', 'model_extractive.distilbert.transformer.layer.5.output_layer_norm.bias', 'model_extractive.distilbert.transformer.layer.5.output_layer_norm.weight', 'model_extractive.distilbert.transformer.layer.5.sa_layer_norm.bias', 'model_extractive.distilbert.transformer.layer.5.sa_layer_norm.weight', 'model_extractive.qa_outputs.bias', 'model_extractive.qa_outputs.weight', 'model_generative.decoder.block.0.layer.0.SelfAttention.k.weight', 'model_generative.decoder.block.0.layer.0.SelfAttention.o.weight', 'model_generative.decoder.block.0.layer.0.SelfAttention.q.weight', 'model_generative.decoder.block.0.layer.0.SelfAttention.relative_attention_bias.weight', 'model_generative.decoder.block.0.layer.0.SelfAttention.v.weight', 'model_generative.decoder.block.0.layer.0.layer_norm.weight', 'model_generative.decoder.block.0.layer.1.EncDecAttention.k.weight', 'model_generative.decoder.block.0.layer.1.EncDecAttention.o.weight', 'model_generative.decoder.block.0.layer.1.EncDecAttention.q.weight', 'model_generative.decoder.block.0.layer.1.EncDecAttention.v.weight', 'model_generative.decoder.block.0.layer.1.layer_norm.weight', 'model_generative.decoder.block.0.layer.2.DenseReluDense.wi.weight', 'model_generative.decoder.block.0.layer.2.DenseReluDense.wo.weight', 'model_generative.decoder.block.0.layer.2.layer_norm.weight', 'model_generative.decoder.block.1.layer.0.SelfAttention.k.weight', 'model_generative.decoder.block.1.layer.0.SelfAttention.o.weight', 'model_generative.decoder.block.1.layer.0.SelfAttention.q.weight', 'model_generative.decoder.block.1.layer.0.SelfAttention.v.weight', 'model_generative.decoder.block.1.layer.0.layer_norm.weight', 'model_generative.decoder.block.1.layer.1.EncDecAttention.k.weight', 'model_generative.decoder.block.1.layer.1.EncDecAttention.o.weight', 'model_generative.decoder.block.1.layer.1.EncDecAttention.q.weight', 'model_generative.decoder.block.1.layer.1.EncDecAttention.v.weight', 'model_generative.decoder.block.1.layer.1.layer_norm.weight', 'model_generative.decoder.block.1.layer.2.DenseReluDense.wi.weight', 'model_generative.decoder.block.1.layer.2.DenseReluDense.wo.weight', 'model_generative.decoder.block.1.layer.2.layer_norm.weight', 'model_generative.decoder.block.10.layer.0.SelfAttention.k.weight', 'model_generative.decoder.block.10.layer.0.SelfAttention.o.weight', 'model_generative.decoder.block.10.layer.0.SelfAttention.q.weight', 'model_generative.decoder.block.10.layer.0.SelfAttention.v.weight', 'model_generative.decoder.block.10.layer.0.layer_norm.weight', 'model_generative.decoder.block.10.layer.1.EncDecAttention.k.weight', 'model_generative.decoder.block.10.layer.1.EncDecAttention.o.weight', 'model_generative.decoder.block.10.layer.1.EncDecAttention.q.weight', 'model_generative.decoder.block.10.layer.1.EncDecAttention.v.weight', 'model_generative.decoder.block.10.layer.1.layer_norm.weight', 'model_generative.decoder.block.10.layer.2.DenseReluDense.wi.weight', 'model_generative.decoder.block.10.layer.2.DenseReluDense.wo.weight', 'model_generative.decoder.block.10.layer.2.layer_norm.weight', 'model_generative.decoder.block.11.layer.0.SelfAttention.k.weight', 'model_generative.decoder.block.11.layer.0.SelfAttention.o.weight', 'model_generative.decoder.block.11.layer.0.SelfAttention.q.weight', 'model_generative.decoder.block.11.layer.0.SelfAttention.v.weight', 'model_generative.decoder.block.11.layer.0.layer_norm.weight', 'model_generative.decoder.block.11.layer.1.EncDecAttention.k.weight', 'model_generative.decoder.block.11.layer.1.EncDecAttention.o.weight', 'model_generative.decoder.block.11.layer.1.EncDecAttention.q.weight', 'model_generative.decoder.block.11.layer.1.EncDecAttention.v.weight', 'model_generative.decoder.block.11.layer.1.layer_norm.weight', 'model_generative.decoder.block.11.layer.2.DenseReluDense.wi.weight', 'model_generative.decoder.block.11.layer.2.DenseReluDense.wo.weight', 'model_generative.decoder.block.11.layer.2.layer_norm.weight', 'model_generative.decoder.block.2.layer.0.SelfAttention.k.weight', 'model_generative.decoder.block.2.layer.0.SelfAttention.o.weight', 'model_generative.decoder.block.2.layer.0.SelfAttention.q.weight', 'model_generative.decoder.block.2.layer.0.SelfAttention.v.weight', 'model_generative.decoder.block.2.layer.0.layer_norm.weight', 'model_generative.decoder.block.2.layer.1.EncDecAttention.k.weight', 'model_generative.decoder.block.2.layer.1.EncDecAttention.o.weight', 'model_generative.decoder.block.2.layer.1.EncDecAttention.q.weight', 'model_generative.decoder.block.2.layer.1.EncDecAttention.v.weight', 'model_generative.decoder.block.2.layer.1.layer_norm.weight', 'model_generative.decoder.block.2.layer.2.DenseReluDense.wi.weight', 'model_generative.decoder.block.2.layer.2.DenseReluDense.wo.weight', 'model_generative.decoder.block.2.layer.2.layer_norm.weight', 'model_generative.decoder.block.3.layer.0.SelfAttention.k.weight', 'model_generative.decoder.block.3.layer.0.SelfAttention.o.weight', 'model_generative.decoder.block.3.layer.0.SelfAttention.q.weight', 'model_generative.decoder.block.3.layer.0.SelfAttention.v.weight', 'model_generative.decoder.block.3.layer.0.layer_norm.weight', 'model_generative.decoder.block.3.layer.1.EncDecAttention.k.weight', 'model_generative.decoder.block.3.layer.1.EncDecAttention.o.weight', 'model_generative.decoder.block.3.layer.1.EncDecAttention.q.weight', 'model_generative.decoder.block.3.layer.1.EncDecAttention.v.weight', 'model_generative.decoder.block.3.layer.1.layer_norm.weight', 'model_generative.decoder.block.3.layer.2.DenseReluDense.wi.weight', 'model_generative.decoder.block.3.layer.2.DenseReluDense.wo.weight', 'model_generative.decoder.block.3.layer.2.layer_norm.weight', 'model_generative.decoder.block.4.layer.0.SelfAttention.k.weight', 'model_generative.decoder.block.4.layer.0.SelfAttention.o.weight', 'model_generative.decoder.block.4.layer.0.SelfAttention.q.weight', 'model_generative.decoder.block.4.layer.0.SelfAttention.v.weight', 'model_generative.decoder.block.4.layer.0.layer_norm.weight', 'model_generative.decoder.block.4.layer.1.EncDecAttention.k.weight', 'model_generative.decoder.block.4.layer.1.EncDecAttention.o.weight', 'model_generative.decoder.block.4.layer.1.EncDecAttention.q.weight', 'model_generative.decoder.block.4.layer.1.EncDecAttention.v.weight', 'model_generative.decoder.block.4.layer.1.layer_norm.weight', 'model_generative.decoder.block.4.layer.2.DenseReluDense.wi.weight', 'model_generative.decoder.block.4.layer.2.DenseReluDense.wo.weight', 'model_generative.decoder.block.4.layer.2.layer_norm.weight', 'model_generative.decoder.block.5.layer.0.SelfAttention.k.weight', 'model_generative.decoder.block.5.layer.0.SelfAttention.o.weight', 'model_generative.decoder.block.5.layer.0.SelfAttention.q.weight', 'model_generative.decoder.block.5.layer.0.SelfAttention.v.weight', 'model_generative.decoder.block.5.layer.0.layer_norm.weight', 'model_generative.decoder.block.5.layer.1.EncDecAttention.k.weight', 'model_generative.decoder.block.5.layer.1.EncDecAttention.o.weight', 'model_generative.decoder.block.5.layer.1.EncDecAttention.q.weight', 'model_generative.decoder.block.5.layer.1.EncDecAttention.v.weight', 'model_generative.decoder.block.5.layer.1.layer_norm.weight', 'model_generative.decoder.block.5.layer.2.DenseReluDense.wi.weight', 'model_generative.decoder.block.5.layer.2.DenseReluDense.wo.weight', 'model_generative.decoder.block.5.layer.2.layer_norm.weight', 'model_generative.decoder.block.6.layer.0.SelfAttention.k.weight', 'model_generative.decoder.block.6.layer.0.SelfAttention.o.weight', 'model_generative.decoder.block.6.layer.0.SelfAttention.q.weight', 'model_generative.decoder.block.6.layer.0.SelfAttention.v.weight', 'model_generative.decoder.block.6.layer.0.layer_norm.weight', 'model_generative.decoder.block.6.layer.1.EncDecAttention.k.weight', 'model_generative.decoder.block.6.layer.1.EncDecAttention.o.weight', 'model_generative.decoder.block.6.layer.1.EncDecAttention.q.weight', 'model_generative.decoder.block.6.layer.1.EncDecAttention.v.weight', 'model_generative.decoder.block.6.layer.1.layer_norm.weight', 'model_generative.decoder.block.6.layer.2.DenseReluDense.wi.weight', 'model_generative.decoder.block.6.layer.2.DenseReluDense.wo.weight', 'model_generative.decoder.block.6.layer.2.layer_norm.weight', 'model_generative.decoder.block.7.layer.0.SelfAttention.k.weight', 'model_generative.decoder.block.7.layer.0.SelfAttention.o.weight', 'model_generative.decoder.block.7.layer.0.SelfAttention.q.weight', 'model_generative.decoder.block.7.layer.0.SelfAttention.v.weight', 'model_generative.decoder.block.7.layer.0.layer_norm.weight', 'model_generative.decoder.block.7.layer.1.EncDecAttention.k.weight', 'model_generative.decoder.block.7.layer.1.EncDecAttention.o.weight', 'model_generative.decoder.block.7.layer.1.EncDecAttention.q.weight', 'model_generative.decoder.block.7.layer.1.EncDecAttention.v.weight', 'model_generative.decoder.block.7.layer.1.layer_norm.weight', 'model_generative.decoder.block.7.layer.2.DenseReluDense.wi.weight', 'model_generative.decoder.block.7.layer.2.DenseReluDense.wo.weight', 'model_generative.decoder.block.7.layer.2.layer_norm.weight', 'model_generative.decoder.block.8.layer.0.SelfAttention.k.weight', 'model_generative.decoder.block.8.layer.0.SelfAttention.o.weight', 'model_generative.decoder.block.8.layer.0.SelfAttention.q.weight', 'model_generative.decoder.block.8.layer.0.SelfAttention.v.weight', 'model_generative.decoder.block.8.layer.0.layer_norm.weight', 'model_generative.decoder.block.8.layer.1.EncDecAttention.k.weight', 'model_generative.decoder.block.8.layer.1.EncDecAttention.o.weight', 'model_generative.decoder.block.8.layer.1.EncDecAttention.q.weight', 'model_generative.decoder.block.8.layer.1.EncDecAttention.v.weight', 'model_generative.decoder.block.8.layer.1.layer_norm.weight', 'model_generative.decoder.block.8.layer.2.DenseReluDense.wi.weight', 'model_generative.decoder.block.8.layer.2.DenseReluDense.wo.weight', 'model_generative.decoder.block.8.layer.2.layer_norm.weight', 'model_generative.decoder.block.9.layer.0.SelfAttention.k.weight', 'model_generative.decoder.block.9.layer.0.SelfAttention.o.weight', 'model_generative.decoder.block.9.layer.0.SelfAttention.q.weight', 'model_generative.decoder.block.9.layer.0.SelfAttention.v.weight', 'model_generative.decoder.block.9.layer.0.layer_norm.weight', 'model_generative.decoder.block.9.layer.1.EncDecAttention.k.weight', 'model_generative.decoder.block.9.layer.1.EncDecAttention.o.weight', 'model_generative.decoder.block.9.layer.1.EncDecAttention.q.weight', 'model_generative.decoder.block.9.layer.1.EncDecAttention.v.weight', 'model_generative.decoder.block.9.layer.1.layer_norm.weight', 'model_generative.decoder.block.9.layer.2.DenseReluDense.wi.weight', 'model_generative.decoder.block.9.layer.2.DenseReluDense.wo.weight', 'model_generative.decoder.block.9.layer.2.layer_norm.weight', 'model_generative.decoder.embed_tokens.weight', 'model_generative.decoder.final_layer_norm.weight', 'model_generative.encoder.block.0.layer.0.SelfAttention.k.weight', 'model_generative.encoder.block.0.layer.0.SelfAttention.o.weight', 'model_generative.encoder.block.0.layer.0.SelfAttention.q.weight', 'model_generative.encoder.block.0.layer.0.SelfAttention.relative_attention_bias.weight', 'model_generative.encoder.block.0.layer.0.SelfAttention.v.weight', 'model_generative.encoder.block.0.layer.0.layer_norm.weight', 'model_generative.encoder.block.0.layer.1.DenseReluDense.wi.weight', 'model_generative.encoder.block.0.layer.1.DenseReluDense.wo.weight', 'model_generative.encoder.block.0.layer.1.layer_norm.weight', 'model_generative.encoder.block.1.layer.0.SelfAttention.k.weight', 'model_generative.encoder.block.1.layer.0.SelfAttention.o.weight', 'model_generative.encoder.block.1.layer.0.SelfAttention.q.weight', 'model_generative.encoder.block.1.layer.0.SelfAttention.v.weight', 'model_generative.encoder.block.1.layer.0.layer_norm.weight', 'model_generative.encoder.block.1.layer.1.DenseReluDense.wi.weight', 'model_generative.encoder.block.1.layer.1.DenseReluDense.wo.weight', 'model_generative.encoder.block.1.layer.1.layer_norm.weight', 'model_generative.encoder.block.10.layer.0.SelfAttention.k.weight', 'model_generative.encoder.block.10.layer.0.SelfAttention.o.weight', 'model_generative.encoder.block.10.layer.0.SelfAttention.q.weight', 'model_generative.encoder.block.10.layer.0.SelfAttention.v.weight', 'model_generative.encoder.block.10.layer.0.layer_norm.weight', 'model_generative.encoder.block.10.layer.1.DenseReluDense.wi.weight', 'model_generative.encoder.block.10.layer.1.DenseReluDense.wo.weight', 'model_generative.encoder.block.10.layer.1.layer_norm.weight', 'model_generative.encoder.block.11.layer.0.SelfAttention.k.weight', 'model_generative.encoder.block.11.layer.0.SelfAttention.o.weight', 'model_generative.encoder.block.11.layer.0.SelfAttention.q.weight', 'model_generative.encoder.block.11.layer.0.SelfAttention.v.weight', 'model_generative.encoder.block.11.layer.0.layer_norm.weight', 'model_generative.encoder.block.11.layer.1.DenseReluDense.wi.weight', 'model_generative.encoder.block.11.layer.1.DenseReluDense.wo.weight', 'model_generative.encoder.block.11.layer.1.layer_norm.weight', 'model_generative.encoder.block.2.layer.0.SelfAttention.k.weight', 'model_generative.encoder.block.2.layer.0.SelfAttention.o.weight', 'model_generative.encoder.block.2.layer.0.SelfAttention.q.weight', 'model_generative.encoder.block.2.layer.0.SelfAttention.v.weight', 'model_generative.encoder.block.2.layer.0.layer_norm.weight', 'model_generative.encoder.block.2.layer.1.DenseReluDense.wi.weight', 'model_generative.encoder.block.2.layer.1.DenseReluDense.wo.weight', 'model_generative.encoder.block.2.layer.1.layer_norm.weight', 'model_generative.encoder.block.3.layer.0.SelfAttention.k.weight', 'model_generative.encoder.block.3.layer.0.SelfAttention.o.weight', 'model_generative.encoder.block.3.layer.0.SelfAttention.q.weight', 'model_generative.encoder.block.3.layer.0.SelfAttention.v.weight', 'model_generative.encoder.block.3.layer.0.layer_norm.weight', 'model_generative.encoder.block.3.layer.1.DenseReluDense.wi.weight', 'model_generative.encoder.block.3.layer.1.DenseReluDense.wo.weight', 'model_generative.encoder.block.3.layer.1.layer_norm.weight', 'model_generative.encoder.block.4.layer.0.SelfAttention.k.weight', 'model_generative.encoder.block.4.layer.0.SelfAttention.o.weight', 'model_generative.encoder.block.4.layer.0.SelfAttention.q.weight', 'model_generative.encoder.block.4.layer.0.SelfAttention.v.weight', 'model_generative.encoder.block.4.layer.0.layer_norm.weight', 'model_generative.encoder.block.4.layer.1.DenseReluDense.wi.weight', 'model_generative.encoder.block.4.layer.1.DenseReluDense.wo.weight', 'model_generative.encoder.block.4.layer.1.layer_norm.weight', 'model_generative.encoder.block.5.layer.0.SelfAttention.k.weight', 'model_generative.encoder.block.5.layer.0.SelfAttention.o.weight', 'model_generative.encoder.block.5.layer.0.SelfAttention.q.weight', 'model_generative.encoder.block.5.layer.0.SelfAttention.v.weight', 'model_generative.encoder.block.5.layer.0.layer_norm.weight', 'model_generative.encoder.block.5.layer.1.DenseReluDense.wi.weight', 'model_generative.encoder.block.5.layer.1.DenseReluDense.wo.weight', 'model_generative.encoder.block.5.layer.1.layer_norm.weight', 'model_generative.encoder.block.6.layer.0.SelfAttention.k.weight', 'model_generative.encoder.block.6.layer.0.SelfAttention.o.weight', 'model_generative.encoder.block.6.layer.0.SelfAttention.q.weight', 'model_generative.encoder.block.6.layer.0.SelfAttention.v.weight', 'model_generative.encoder.block.6.layer.0.layer_norm.weight', 'model_generative.encoder.block.6.layer.1.DenseReluDense.wi.weight', 'model_generative.encoder.block.6.layer.1.DenseReluDense.wo.weight', 'model_generative.encoder.block.6.layer.1.layer_norm.weight', 'model_generative.encoder.block.7.layer.0.SelfAttention.k.weight', 'model_generative.encoder.block.7.layer.0.SelfAttention.o.weight', 'model_generative.encoder.block.7.layer.0.SelfAttention.q.weight', 'model_generative.encoder.block.7.layer.0.SelfAttention.v.weight', 'model_generative.encoder.block.7.layer.0.layer_norm.weight', 'model_generative.encoder.block.7.layer.1.DenseReluDense.wi.weight', 'model_generative.encoder.block.7.layer.1.DenseReluDense.wo.weight', 'model_generative.encoder.block.7.layer.1.layer_norm.weight', 'model_generative.encoder.block.8.layer.0.SelfAttention.k.weight', 'model_generative.encoder.block.8.layer.0.SelfAttention.o.weight', 'model_generative.encoder.block.8.layer.0.SelfAttention.q.weight', 'model_generative.encoder.block.8.layer.0.SelfAttention.v.weight', 'model_generative.encoder.block.8.layer.0.layer_norm.weight', 'model_generative.encoder.block.8.layer.1.DenseReluDense.wi.weight', 'model_generative.encoder.block.8.layer.1.DenseReluDense.wo.weight', 'model_generative.encoder.block.8.layer.1.layer_norm.weight', 'model_generative.encoder.block.9.layer.0.SelfAttention.k.weight', 'model_generative.encoder.block.9.layer.0.SelfAttention.o.weight', 'model_generative.encoder.block.9.layer.0.SelfAttention.q.weight', 'model_generative.encoder.block.9.layer.0.SelfAttention.v.weight', 'model_generative.encoder.block.9.layer.0.layer_norm.weight', 'model_generative.encoder.block.9.layer.1.DenseReluDense.wi.weight', 'model_generative.encoder.block.9.layer.1.DenseReluDense.wo.weight', 'model_generative.encoder.block.9.layer.1.layer_norm.weight', 'model_generative.encoder.embed_tokens.weight', 'model_generative.encoder.final_layer_norm.weight', 'model_generative.lm_head.weight', 'model_generative.shared.weight']\n",
|
144 |
+
"- This IS expected if you are initializing DistilBertForQuestionAnswering from the checkpoint of a model trained on another task or with another architecture (e.g. initializing a BertForSequenceClassification model from a BertForPreTraining model).\n",
|
145 |
+
"- This IS NOT expected if you are initializing DistilBertForQuestionAnswering from the checkpoint of a model that you expect to be exactly identical (initializing a BertForSequenceClassification model from a BertForSequenceClassification model).\n",
|
146 |
+
"Some weights of DistilBertForQuestionAnswering were not initialized from the model checkpoint at justinhl/hybrid-qa and are newly initialized: ['embeddings.LayerNorm.bias', 'embeddings.LayerNorm.weight', 'embeddings.position_embeddings.weight', 'embeddings.word_embeddings.weight', 'qa_outputs.bias', 'qa_outputs.weight', 'transformer.layer.0.attention.k_lin.bias', 'transformer.layer.0.attention.k_lin.weight', 'transformer.layer.0.attention.out_lin.bias', 'transformer.layer.0.attention.out_lin.weight', 'transformer.layer.0.attention.q_lin.bias', 'transformer.layer.0.attention.q_lin.weight', 'transformer.layer.0.attention.v_lin.bias', 'transformer.layer.0.attention.v_lin.weight', 'transformer.layer.0.ffn.lin1.bias', 'transformer.layer.0.ffn.lin1.weight', 'transformer.layer.0.ffn.lin2.bias', 'transformer.layer.0.ffn.lin2.weight', 'transformer.layer.0.output_layer_norm.bias', 'transformer.layer.0.output_layer_norm.weight', 'transformer.layer.0.sa_layer_norm.bias', 'transformer.layer.0.sa_layer_norm.weight', 'transformer.layer.1.attention.k_lin.bias', 'transformer.layer.1.attention.k_lin.weight', 'transformer.layer.1.attention.out_lin.bias', 'transformer.layer.1.attention.out_lin.weight', 'transformer.layer.1.attention.q_lin.bias', 'transformer.layer.1.attention.q_lin.weight', 'transformer.layer.1.attention.v_lin.bias', 'transformer.layer.1.attention.v_lin.weight', 'transformer.layer.1.ffn.lin1.bias', 'transformer.layer.1.ffn.lin1.weight', 'transformer.layer.1.ffn.lin2.bias', 'transformer.layer.1.ffn.lin2.weight', 'transformer.layer.1.output_layer_norm.bias', 'transformer.layer.1.output_layer_norm.weight', 'transformer.layer.1.sa_layer_norm.bias', 'transformer.layer.1.sa_layer_norm.weight', 'transformer.layer.2.attention.k_lin.bias', 'transformer.layer.2.attention.k_lin.weight', 'transformer.layer.2.attention.out_lin.bias', 'transformer.layer.2.attention.out_lin.weight', 'transformer.layer.2.attention.q_lin.bias', 'transformer.layer.2.attention.q_lin.weight', 'transformer.layer.2.attention.v_lin.bias', 'transformer.layer.2.attention.v_lin.weight', 'transformer.layer.2.ffn.lin1.bias', 'transformer.layer.2.ffn.lin1.weight', 'transformer.layer.2.ffn.lin2.bias', 'transformer.layer.2.ffn.lin2.weight', 'transformer.layer.2.output_layer_norm.bias', 'transformer.layer.2.output_layer_norm.weight', 'transformer.layer.2.sa_layer_norm.bias', 'transformer.layer.2.sa_layer_norm.weight', 'transformer.layer.3.attention.k_lin.bias', 'transformer.layer.3.attention.k_lin.weight', 'transformer.layer.3.attention.out_lin.bias', 'transformer.layer.3.attention.out_lin.weight', 'transformer.layer.3.attention.q_lin.bias', 'transformer.layer.3.attention.q_lin.weight', 'transformer.layer.3.attention.v_lin.bias', 'transformer.layer.3.attention.v_lin.weight', 'transformer.layer.3.ffn.lin1.bias', 'transformer.layer.3.ffn.lin1.weight', 'transformer.layer.3.ffn.lin2.bias', 'transformer.layer.3.ffn.lin2.weight', 'transformer.layer.3.output_layer_norm.bias', 'transformer.layer.3.output_layer_norm.weight', 'transformer.layer.3.sa_layer_norm.bias', 'transformer.layer.3.sa_layer_norm.weight', 'transformer.layer.4.attention.k_lin.bias', 'transformer.layer.4.attention.k_lin.weight', 'transformer.layer.4.attention.out_lin.bias', 'transformer.layer.4.attention.out_lin.weight', 'transformer.layer.4.attention.q_lin.bias', 'transformer.layer.4.attention.q_lin.weight', 'transformer.layer.4.attention.v_lin.bias', 'transformer.layer.4.attention.v_lin.weight', 'transformer.layer.4.ffn.lin1.bias', 'transformer.layer.4.ffn.lin1.weight', 'transformer.layer.4.ffn.lin2.bias', 'transformer.layer.4.ffn.lin2.weight', 'transformer.layer.4.output_layer_norm.bias', 'transformer.layer.4.output_layer_norm.weight', 'transformer.layer.4.sa_layer_norm.bias', 'transformer.layer.4.sa_layer_norm.weight', 'transformer.layer.5.attention.k_lin.bias', 'transformer.layer.5.attention.k_lin.weight', 'transformer.layer.5.attention.out_lin.bias', 'transformer.layer.5.attention.out_lin.weight', 'transformer.layer.5.attention.q_lin.bias', 'transformer.layer.5.attention.q_lin.weight', 'transformer.layer.5.attention.v_lin.bias', 'transformer.layer.5.attention.v_lin.weight', 'transformer.layer.5.ffn.lin1.bias', 'transformer.layer.5.ffn.lin1.weight', 'transformer.layer.5.ffn.lin2.bias', 'transformer.layer.5.ffn.lin2.weight', 'transformer.layer.5.output_layer_norm.bias', 'transformer.layer.5.output_layer_norm.weight', 'transformer.layer.5.sa_layer_norm.bias', 'transformer.layer.5.sa_layer_norm.weight']\n",
|
147 |
+
"You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\n"
|
148 |
+
]
|
149 |
+
}
|
150 |
+
],
|
151 |
+
"source": [
|
152 |
+
"# Importing from remote\n",
|
153 |
+
"imported_pipe = pipeline(\"hybrid-qa\", model=\"justinhl/hybrid-qa\", trust_remote_code=True)"
|
154 |
+
]
|
155 |
+
},
|
156 |
+
{
|
157 |
+
"cell_type": "code",
|
158 |
+
"source": [
|
159 |
+
"# Inference testing!\n",
|
160 |
+
"imported_pipe(question=\"What is the capital of Norway?\",context=\"The capital of Norway is Oslo\")"
|
161 |
+
],
|
162 |
+
"metadata": {
|
163 |
+
"colab": {
|
164 |
+
"base_uri": "https://localhost:8080/"
|
165 |
+
},
|
166 |
+
"id": "sQsoT-UpPp0O",
|
167 |
+
"outputId": "dd922309-bd21-4684-caee-c4d4499bf69b"
|
168 |
+
},
|
169 |
+
"execution_count": 8,
|
170 |
+
"outputs": [
|
171 |
+
{
|
172 |
+
"output_type": "stream",
|
173 |
+
"name": "stderr",
|
174 |
+
"text": [
|
175 |
+
"Truncation was not explicitly activated but `max_length` is provided a specific value, please use `truncation=True` to explicitly truncate examples to max length. Defaulting to 'longest_first' truncation strategy. If you encode pairs of sequences (GLUE-style) with the tokenizer you can select this strategy more precisely by providing a specific strategy to `truncation`.\n",
|
176 |
+
"/usr/local/lib/python3.10/dist-packages/transformers/generation/utils.py:1141: UserWarning: Using the model-agnostic default `max_length` (=20) to control the generation length. We recommend setting `max_new_tokens` to control the maximum length of the generation.\n",
|
177 |
+
" warnings.warn(\n"
|
178 |
+
]
|
179 |
+
},
|
180 |
+
{
|
181 |
+
"output_type": "execute_result",
|
182 |
+
"data": {
|
183 |
+
"text/plain": [
|
184 |
+
"{'guess': 'Oslo', 'confidence': 2.0940363768613864e-14}"
|
185 |
+
]
|
186 |
+
},
|
187 |
+
"metadata": {},
|
188 |
+
"execution_count": 8
|
189 |
+
}
|
190 |
+
]
|
191 |
+
},
|
192 |
+
{
|
193 |
+
"cell_type": "code",
|
194 |
+
"source": [
|
195 |
+
"print(\"Model loaded:\", imported_pipe.model)"
|
196 |
+
],
|
197 |
+
"metadata": {
|
198 |
+
"colab": {
|
199 |
+
"base_uri": "https://localhost:8080/"
|
200 |
+
},
|
201 |
+
"id": "GEmtld6OVT7W",
|
202 |
+
"outputId": "93217a25-668e-4a46-8fc9-9db440693a1c"
|
203 |
+
},
|
204 |
+
"execution_count": 9,
|
205 |
+
"outputs": [
|
206 |
+
{
|
207 |
+
"output_type": "stream",
|
208 |
+
"name": "stdout",
|
209 |
+
"text": [
|
210 |
+
"Model loaded: HybridQAModel(\n",
|
211 |
+
" (model_extractive): DistilBertForQuestionAnswering(\n",
|
212 |
+
" (distilbert): DistilBertModel(\n",
|
213 |
+
" (embeddings): Embeddings(\n",
|
214 |
+
" (word_embeddings): Embedding(30522, 768, padding_idx=0)\n",
|
215 |
+
" (position_embeddings): Embedding(512, 768)\n",
|
216 |
+
" (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
|
217 |
+
" (dropout): Dropout(p=0.1, inplace=False)\n",
|
218 |
+
" )\n",
|
219 |
+
" (transformer): Transformer(\n",
|
220 |
+
" (layer): ModuleList(\n",
|
221 |
+
" (0-5): 6 x TransformerBlock(\n",
|
222 |
+
" (attention): MultiHeadSelfAttention(\n",
|
223 |
+
" (dropout): Dropout(p=0.1, inplace=False)\n",
|
224 |
+
" (q_lin): Linear(in_features=768, out_features=768, bias=True)\n",
|
225 |
+
" (k_lin): Linear(in_features=768, out_features=768, bias=True)\n",
|
226 |
+
" (v_lin): Linear(in_features=768, out_features=768, bias=True)\n",
|
227 |
+
" (out_lin): Linear(in_features=768, out_features=768, bias=True)\n",
|
228 |
+
" )\n",
|
229 |
+
" (sa_layer_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
|
230 |
+
" (ffn): FFN(\n",
|
231 |
+
" (dropout): Dropout(p=0.1, inplace=False)\n",
|
232 |
+
" (lin1): Linear(in_features=768, out_features=3072, bias=True)\n",
|
233 |
+
" (lin2): Linear(in_features=3072, out_features=768, bias=True)\n",
|
234 |
+
" (activation): GELUActivation()\n",
|
235 |
+
" )\n",
|
236 |
+
" (output_layer_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
|
237 |
+
" )\n",
|
238 |
+
" )\n",
|
239 |
+
" )\n",
|
240 |
+
" )\n",
|
241 |
+
" (qa_outputs): Linear(in_features=768, out_features=2, bias=True)\n",
|
242 |
+
" (dropout): Dropout(p=0.1, inplace=False)\n",
|
243 |
+
" )\n",
|
244 |
+
" (model_generative): T5ForConditionalGeneration(\n",
|
245 |
+
" (shared): Embedding(32128, 768)\n",
|
246 |
+
" (encoder): T5Stack(\n",
|
247 |
+
" (embed_tokens): Embedding(32128, 768)\n",
|
248 |
+
" (block): ModuleList(\n",
|
249 |
+
" (0): T5Block(\n",
|
250 |
+
" (layer): ModuleList(\n",
|
251 |
+
" (0): T5LayerSelfAttention(\n",
|
252 |
+
" (SelfAttention): T5Attention(\n",
|
253 |
+
" (q): Linear(in_features=768, out_features=768, bias=False)\n",
|
254 |
+
" (k): Linear(in_features=768, out_features=768, bias=False)\n",
|
255 |
+
" (v): Linear(in_features=768, out_features=768, bias=False)\n",
|
256 |
+
" (o): Linear(in_features=768, out_features=768, bias=False)\n",
|
257 |
+
" (relative_attention_bias): Embedding(32, 12)\n",
|
258 |
+
" )\n",
|
259 |
+
" (layer_norm): T5LayerNorm()\n",
|
260 |
+
" (dropout): Dropout(p=0.1, inplace=False)\n",
|
261 |
+
" )\n",
|
262 |
+
" (1): T5LayerFF(\n",
|
263 |
+
" (DenseReluDense): T5DenseActDense(\n",
|
264 |
+
" (wi): Linear(in_features=768, out_features=3072, bias=False)\n",
|
265 |
+
" (wo): Linear(in_features=3072, out_features=768, bias=False)\n",
|
266 |
+
" (dropout): Dropout(p=0.1, inplace=False)\n",
|
267 |
+
" (act): ReLU()\n",
|
268 |
+
" )\n",
|
269 |
+
" (layer_norm): T5LayerNorm()\n",
|
270 |
+
" (dropout): Dropout(p=0.1, inplace=False)\n",
|
271 |
+
" )\n",
|
272 |
+
" )\n",
|
273 |
+
" )\n",
|
274 |
+
" (1-11): 11 x T5Block(\n",
|
275 |
+
" (layer): ModuleList(\n",
|
276 |
+
" (0): T5LayerSelfAttention(\n",
|
277 |
+
" (SelfAttention): T5Attention(\n",
|
278 |
+
" (q): Linear(in_features=768, out_features=768, bias=False)\n",
|
279 |
+
" (k): Linear(in_features=768, out_features=768, bias=False)\n",
|
280 |
+
" (v): Linear(in_features=768, out_features=768, bias=False)\n",
|
281 |
+
" (o): Linear(in_features=768, out_features=768, bias=False)\n",
|
282 |
+
" )\n",
|
283 |
+
" (layer_norm): T5LayerNorm()\n",
|
284 |
+
" (dropout): Dropout(p=0.1, inplace=False)\n",
|
285 |
+
" )\n",
|
286 |
+
" (1): T5LayerFF(\n",
|
287 |
+
" (DenseReluDense): T5DenseActDense(\n",
|
288 |
+
" (wi): Linear(in_features=768, out_features=3072, bias=False)\n",
|
289 |
+
" (wo): Linear(in_features=3072, out_features=768, bias=False)\n",
|
290 |
+
" (dropout): Dropout(p=0.1, inplace=False)\n",
|
291 |
+
" (act): ReLU()\n",
|
292 |
+
" )\n",
|
293 |
+
" (layer_norm): T5LayerNorm()\n",
|
294 |
+
" (dropout): Dropout(p=0.1, inplace=False)\n",
|
295 |
+
" )\n",
|
296 |
+
" )\n",
|
297 |
+
" )\n",
|
298 |
+
" )\n",
|
299 |
+
" (final_layer_norm): T5LayerNorm()\n",
|
300 |
+
" (dropout): Dropout(p=0.1, inplace=False)\n",
|
301 |
+
" )\n",
|
302 |
+
" (decoder): T5Stack(\n",
|
303 |
+
" (embed_tokens): Embedding(32128, 768)\n",
|
304 |
+
" (block): ModuleList(\n",
|
305 |
+
" (0): T5Block(\n",
|
306 |
+
" (layer): ModuleList(\n",
|
307 |
+
" (0): T5LayerSelfAttention(\n",
|
308 |
+
" (SelfAttention): T5Attention(\n",
|
309 |
+
" (q): Linear(in_features=768, out_features=768, bias=False)\n",
|
310 |
+
" (k): Linear(in_features=768, out_features=768, bias=False)\n",
|
311 |
+
" (v): Linear(in_features=768, out_features=768, bias=False)\n",
|
312 |
+
" (o): Linear(in_features=768, out_features=768, bias=False)\n",
|
313 |
+
" (relative_attention_bias): Embedding(32, 12)\n",
|
314 |
+
" )\n",
|
315 |
+
" (layer_norm): T5LayerNorm()\n",
|
316 |
+
" (dropout): Dropout(p=0.1, inplace=False)\n",
|
317 |
+
" )\n",
|
318 |
+
" (1): T5LayerCrossAttention(\n",
|
319 |
+
" (EncDecAttention): T5Attention(\n",
|
320 |
+
" (q): Linear(in_features=768, out_features=768, bias=False)\n",
|
321 |
+
" (k): Linear(in_features=768, out_features=768, bias=False)\n",
|
322 |
+
" (v): Linear(in_features=768, out_features=768, bias=False)\n",
|
323 |
+
" (o): Linear(in_features=768, out_features=768, bias=False)\n",
|
324 |
+
" )\n",
|
325 |
+
" (layer_norm): T5LayerNorm()\n",
|
326 |
+
" (dropout): Dropout(p=0.1, inplace=False)\n",
|
327 |
+
" )\n",
|
328 |
+
" (2): T5LayerFF(\n",
|
329 |
+
" (DenseReluDense): T5DenseActDense(\n",
|
330 |
+
" (wi): Linear(in_features=768, out_features=3072, bias=False)\n",
|
331 |
+
" (wo): Linear(in_features=3072, out_features=768, bias=False)\n",
|
332 |
+
" (dropout): Dropout(p=0.1, inplace=False)\n",
|
333 |
+
" (act): ReLU()\n",
|
334 |
+
" )\n",
|
335 |
+
" (layer_norm): T5LayerNorm()\n",
|
336 |
+
" (dropout): Dropout(p=0.1, inplace=False)\n",
|
337 |
+
" )\n",
|
338 |
+
" )\n",
|
339 |
+
" )\n",
|
340 |
+
" (1-11): 11 x T5Block(\n",
|
341 |
+
" (layer): ModuleList(\n",
|
342 |
+
" (0): T5LayerSelfAttention(\n",
|
343 |
+
" (SelfAttention): T5Attention(\n",
|
344 |
+
" (q): Linear(in_features=768, out_features=768, bias=False)\n",
|
345 |
+
" (k): Linear(in_features=768, out_features=768, bias=False)\n",
|
346 |
+
" (v): Linear(in_features=768, out_features=768, bias=False)\n",
|
347 |
+
" (o): Linear(in_features=768, out_features=768, bias=False)\n",
|
348 |
+
" )\n",
|
349 |
+
" (layer_norm): T5LayerNorm()\n",
|
350 |
+
" (dropout): Dropout(p=0.1, inplace=False)\n",
|
351 |
+
" )\n",
|
352 |
+
" (1): T5LayerCrossAttention(\n",
|
353 |
+
" (EncDecAttention): T5Attention(\n",
|
354 |
+
" (q): Linear(in_features=768, out_features=768, bias=False)\n",
|
355 |
+
" (k): Linear(in_features=768, out_features=768, bias=False)\n",
|
356 |
+
" (v): Linear(in_features=768, out_features=768, bias=False)\n",
|
357 |
+
" (o): Linear(in_features=768, out_features=768, bias=False)\n",
|
358 |
+
" )\n",
|
359 |
+
" (layer_norm): T5LayerNorm()\n",
|
360 |
+
" (dropout): Dropout(p=0.1, inplace=False)\n",
|
361 |
+
" )\n",
|
362 |
+
" (2): T5LayerFF(\n",
|
363 |
+
" (DenseReluDense): T5DenseActDense(\n",
|
364 |
+
" (wi): Linear(in_features=768, out_features=3072, bias=False)\n",
|
365 |
+
" (wo): Linear(in_features=3072, out_features=768, bias=False)\n",
|
366 |
+
" (dropout): Dropout(p=0.1, inplace=False)\n",
|
367 |
+
" (act): ReLU()\n",
|
368 |
+
" )\n",
|
369 |
+
" (layer_norm): T5LayerNorm()\n",
|
370 |
+
" (dropout): Dropout(p=0.1, inplace=False)\n",
|
371 |
+
" )\n",
|
372 |
+
" )\n",
|
373 |
+
" )\n",
|
374 |
+
" )\n",
|
375 |
+
" (final_layer_norm): T5LayerNorm()\n",
|
376 |
+
" (dropout): Dropout(p=0.1, inplace=False)\n",
|
377 |
+
" )\n",
|
378 |
+
" (lm_head): Linear(in_features=768, out_features=32128, bias=False)\n",
|
379 |
+
" )\n",
|
380 |
+
")\n"
|
381 |
+
]
|
382 |
+
}
|
383 |
+
]
|
384 |
+
}
|
385 |
+
],
|
386 |
+
"metadata": {
|
387 |
+
"kernelspec": {
|
388 |
+
"display_name": "Python 3",
|
389 |
+
"language": "python",
|
390 |
+
"name": "python3"
|
391 |
+
},
|
392 |
+
"language_info": {
|
393 |
+
"codemirror_mode": {
|
394 |
+
"name": "ipython",
|
395 |
+
"version": 3
|
396 |
+
},
|
397 |
+
"file_extension": ".py",
|
398 |
+
"mimetype": "text/x-python",
|
399 |
+
"name": "python",
|
400 |
+
"nbconvert_exporter": "python",
|
401 |
+
"pygments_lexer": "ipython3",
|
402 |
+
"version": "3.11.7"
|
403 |
+
},
|
404 |
+
"colab": {
|
405 |
+
"provenance": []
|
406 |
+
}
|
407 |
+
},
|
408 |
+
"nbformat": 4,
|
409 |
+
"nbformat_minor": 0
|
410 |
+
}
|