Upload tokenizer_info.json with huggingface_hub
Browse files- tokenizer_info.json +59 -0
tokenizer_info.json
ADDED
|
@@ -0,0 +1,59 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"vocab_size": 10303,
|
| 3 |
+
"pad_token_id": 0,
|
| 4 |
+
"mf_start_token_id": 1,
|
| 5 |
+
"mf_end_token_id": 2,
|
| 6 |
+
"bp_start_token_id": 3,
|
| 7 |
+
"bp_end_token_id": 4,
|
| 8 |
+
"cc_start_token_id": 5,
|
| 9 |
+
"cc_end_token_id": 6,
|
| 10 |
+
"organism_vocab_size": 201,
|
| 11 |
+
"embed_model_type": "esm2",
|
| 12 |
+
"max_protein_length": 1024,
|
| 13 |
+
"vocab_config": {
|
| 14 |
+
"MF": {
|
| 15 |
+
"method": "min_freq",
|
| 16 |
+
"value": 20
|
| 17 |
+
},
|
| 18 |
+
"BP": {
|
| 19 |
+
"method": "min_freq",
|
| 20 |
+
"value": 20
|
| 21 |
+
},
|
| 22 |
+
"CC": {
|
| 23 |
+
"method": "min_freq",
|
| 24 |
+
"value": 20
|
| 25 |
+
}
|
| 26 |
+
},
|
| 27 |
+
"vocab_stats": {
|
| 28 |
+
"MF": {
|
| 29 |
+
"method": "min_freq",
|
| 30 |
+
"threshold": 20,
|
| 31 |
+
"selected": 1537,
|
| 32 |
+
"freq_range": [
|
| 33 |
+
20,
|
| 34 |
+
73296
|
| 35 |
+
],
|
| 36 |
+
"total_terms": 7183
|
| 37 |
+
},
|
| 38 |
+
"BP": {
|
| 39 |
+
"method": "min_freq",
|
| 40 |
+
"threshold": 20,
|
| 41 |
+
"selected": 7720,
|
| 42 |
+
"freq_range": [
|
| 43 |
+
20,
|
| 44 |
+
81287
|
| 45 |
+
],
|
| 46 |
+
"total_terms": 21200
|
| 47 |
+
},
|
| 48 |
+
"CC": {
|
| 49 |
+
"method": "min_freq",
|
| 50 |
+
"threshold": 20,
|
| 51 |
+
"selected": 1039,
|
| 52 |
+
"freq_range": [
|
| 53 |
+
20,
|
| 54 |
+
84240
|
| 55 |
+
],
|
| 56 |
+
"total_terms": 2939
|
| 57 |
+
}
|
| 58 |
+
}
|
| 59 |
+
}
|