eihab2342/code-efficiency
0
1{
2 "_name_or_path": "Salesforce/codet5-small",
3 "architectures": [
4 "T5ForConditionalGeneration"
5 ],
6 "bos_token_id": 1,
7 "classifier_dropout": 0.0,
8 "d_ff": 2048,
9 "d_kv": 64,
10 "d_model": 512,
11 "decoder_start_token_id": 0,
12 "dense_act_fn": "relu",
13 "dropout_rate": 0.1,
14 "eos_token_id": 2,
15 "feed_forward_proj": "relu",
16 "gradient_checkpointing": false,
17 "id2label": {
18 "0": "LABEL_0"
19 },
20 "initializer_factor": 1.0,
21 "is_encoder_decoder": true,
22 "is_gated_act": false,
23 "label2id": {
24 "LABEL_0": 0
25 },
26 "layer_norm_epsilon": 1e-06,
27 "model_type": "t5",
28 "n_positions": 512,
29 "num_decoder_layers": 6,
30 "num_heads": 8,
31 "num_layers": 6,
32 "output_past": true,
33 "pad_token_id": 0,
34 "relative_attention_max_distance": 128,
35 "relative_attention_num_buckets": 32,
36 "task_specific_params": {
37 "summarization": {
38 "early_stopping": true,
39 "length_penalty": 2.0,
40 "max_length": 200,
41 "min_length": 30,
42 "no_repeat_ngram_size": 3,
43 "num_beams": 4,
44 "prefix": "summarize: "
45 },
46 "translation_en_to_de": {
47 "early_stopping": true,
48 "max_length": 300,
49 "num_beams": 4,
50 "prefix": "translate English to German: "
51 },
52 "translation_en_to_fr": {
53 "early_stopping": true,
54 "max_length": 300,
55 "num_beams": 4,
56 "prefix": "translate English to French: "
57 },
58 "translation_en_to_ro": {
59 "early_stopping": true,
60 "max_length": 300,
61 "num_beams": 4,
62 "prefix": "translate English to Romanian: "
63 }
64 },
65 "torch_dtype": "float32",
66 "transformers_version": "4.36.0",
67 "use_cache": true,
68 "vocab_size": 32100
69}
70 