| { |
| "format": "shared-character-private-continuation-source-v1", |
| "parameters": 74842752, |
| "corrections": [ |
| { |
| "key": "t000", |
| "name": "embeddings.cls_token", |
| "shape": [ |
| 1, |
| 1, |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t001", |
| "name": "embeddings.distillation_token", |
| "shape": [ |
| 1, |
| 1, |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t002", |
| "name": "embeddings.patch_embeddings.projection.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t003", |
| "name": "embeddings.patch_embeddings.projection.weight", |
| "shape": [ |
| 384, |
| 3, |
| 16, |
| 16 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 768 |
| ] |
| }, |
| { |
| "key": "t004", |
| "name": "embeddings.position_embeddings", |
| "shape": [ |
| 1, |
| 578, |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t005", |
| "name": "encoder.layer.0.attention.attention.key.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t006", |
| "name": "encoder.layer.0.attention.attention.key.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t007", |
| "name": "encoder.layer.0.attention.attention.query.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t008", |
| "name": "encoder.layer.0.attention.attention.query.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t009", |
| "name": "encoder.layer.0.attention.attention.value.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t010", |
| "name": "encoder.layer.0.attention.attention.value.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t011", |
| "name": "encoder.layer.0.attention.output.dense.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t012", |
| "name": "encoder.layer.0.attention.output.dense.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t013", |
| "name": "encoder.layer.0.intermediate.dense.bias", |
| "shape": [ |
| 1536 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t014", |
| "name": "encoder.layer.0.intermediate.dense.weight", |
| "shape": [ |
| 1536, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 1536, |
| 384 |
| ] |
| }, |
| { |
| "key": "t015", |
| "name": "encoder.layer.0.layernorm_after.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t016", |
| "name": "encoder.layer.0.layernorm_after.weight", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t017", |
| "name": "encoder.layer.0.layernorm_before.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t018", |
| "name": "encoder.layer.0.layernorm_before.weight", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t019", |
| "name": "encoder.layer.0.output.dense.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t020", |
| "name": "encoder.layer.0.output.dense.weight", |
| "shape": [ |
| 384, |
| 1536 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 1536 |
| ] |
| }, |
| { |
| "key": "t021", |
| "name": "encoder.layer.1.attention.attention.key.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t022", |
| "name": "encoder.layer.1.attention.attention.key.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t023", |
| "name": "encoder.layer.1.attention.attention.query.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t024", |
| "name": "encoder.layer.1.attention.attention.query.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t025", |
| "name": "encoder.layer.1.attention.attention.value.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t026", |
| "name": "encoder.layer.1.attention.attention.value.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t027", |
| "name": "encoder.layer.1.attention.output.dense.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t028", |
| "name": "encoder.layer.1.attention.output.dense.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t029", |
| "name": "encoder.layer.1.intermediate.dense.bias", |
| "shape": [ |
| 1536 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t030", |
| "name": "encoder.layer.1.intermediate.dense.weight", |
| "shape": [ |
| 1536, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 1536, |
| 384 |
| ] |
| }, |
| { |
| "key": "t031", |
| "name": "encoder.layer.1.layernorm_after.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t032", |
| "name": "encoder.layer.1.layernorm_after.weight", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t033", |
| "name": "encoder.layer.1.layernorm_before.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t034", |
| "name": "encoder.layer.1.layernorm_before.weight", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t035", |
| "name": "encoder.layer.1.output.dense.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t036", |
| "name": "encoder.layer.1.output.dense.weight", |
| "shape": [ |
| 384, |
| 1536 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 1536 |
| ] |
| }, |
| { |
| "key": "t037", |
| "name": "encoder.layer.10.attention.attention.key.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t038", |
| "name": "encoder.layer.10.attention.attention.key.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t039", |
| "name": "encoder.layer.10.attention.attention.query.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t040", |
| "name": "encoder.layer.10.attention.attention.query.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t041", |
| "name": "encoder.layer.10.attention.attention.value.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t042", |
| "name": "encoder.layer.10.attention.attention.value.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t043", |
| "name": "encoder.layer.10.attention.output.dense.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t044", |
| "name": "encoder.layer.10.attention.output.dense.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t045", |
| "name": "encoder.layer.10.intermediate.dense.bias", |
| "shape": [ |
| 1536 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t046", |
| "name": "encoder.layer.10.intermediate.dense.weight", |
| "shape": [ |
| 1536, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 1536, |
| 384 |
| ] |
| }, |
| { |
| "key": "t047", |
| "name": "encoder.layer.10.layernorm_after.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t048", |
| "name": "encoder.layer.10.layernorm_after.weight", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t049", |
| "name": "encoder.layer.10.layernorm_before.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t050", |
| "name": "encoder.layer.10.layernorm_before.weight", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t051", |
| "name": "encoder.layer.10.output.dense.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t052", |
| "name": "encoder.layer.10.output.dense.weight", |
| "shape": [ |
| 384, |
| 1536 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 1536 |
| ] |
| }, |
| { |
| "key": "t053", |
| "name": "encoder.layer.11.attention.attention.key.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t054", |
| "name": "encoder.layer.11.attention.attention.key.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t055", |
| "name": "encoder.layer.11.attention.attention.query.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t056", |
| "name": "encoder.layer.11.attention.attention.query.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t057", |
| "name": "encoder.layer.11.attention.attention.value.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t058", |
| "name": "encoder.layer.11.attention.attention.value.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t059", |
| "name": "encoder.layer.11.attention.output.dense.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t060", |
| "name": "encoder.layer.11.attention.output.dense.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t061", |
| "name": "encoder.layer.11.intermediate.dense.bias", |
| "shape": [ |
| 1536 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t062", |
| "name": "encoder.layer.11.intermediate.dense.weight", |
| "shape": [ |
| 1536, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 1536, |
| 384 |
| ] |
| }, |
| { |
| "key": "t063", |
| "name": "encoder.layer.11.layernorm_after.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t064", |
| "name": "encoder.layer.11.layernorm_after.weight", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t065", |
| "name": "encoder.layer.11.layernorm_before.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t066", |
| "name": "encoder.layer.11.layernorm_before.weight", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t067", |
| "name": "encoder.layer.11.output.dense.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t068", |
| "name": "encoder.layer.11.output.dense.weight", |
| "shape": [ |
| 384, |
| 1536 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 1536 |
| ] |
| }, |
| { |
| "key": "t069", |
| "name": "encoder.layer.2.attention.attention.key.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t070", |
| "name": "encoder.layer.2.attention.attention.key.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t071", |
| "name": "encoder.layer.2.attention.attention.query.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t072", |
| "name": "encoder.layer.2.attention.attention.query.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t073", |
| "name": "encoder.layer.2.attention.attention.value.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t074", |
| "name": "encoder.layer.2.attention.attention.value.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t075", |
| "name": "encoder.layer.2.attention.output.dense.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t076", |
| "name": "encoder.layer.2.attention.output.dense.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t077", |
| "name": "encoder.layer.2.intermediate.dense.bias", |
| "shape": [ |
| 1536 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t078", |
| "name": "encoder.layer.2.intermediate.dense.weight", |
| "shape": [ |
| 1536, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 1536, |
| 384 |
| ] |
| }, |
| { |
| "key": "t079", |
| "name": "encoder.layer.2.layernorm_after.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t080", |
| "name": "encoder.layer.2.layernorm_after.weight", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t081", |
| "name": "encoder.layer.2.layernorm_before.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t082", |
| "name": "encoder.layer.2.layernorm_before.weight", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t083", |
| "name": "encoder.layer.2.output.dense.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t084", |
| "name": "encoder.layer.2.output.dense.weight", |
| "shape": [ |
| 384, |
| 1536 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 1536 |
| ] |
| }, |
| { |
| "key": "t085", |
| "name": "encoder.layer.3.attention.attention.key.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t086", |
| "name": "encoder.layer.3.attention.attention.key.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t087", |
| "name": "encoder.layer.3.attention.attention.query.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t088", |
| "name": "encoder.layer.3.attention.attention.query.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t089", |
| "name": "encoder.layer.3.attention.attention.value.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t090", |
| "name": "encoder.layer.3.attention.attention.value.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t091", |
| "name": "encoder.layer.3.attention.output.dense.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t092", |
| "name": "encoder.layer.3.attention.output.dense.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t093", |
| "name": "encoder.layer.3.intermediate.dense.bias", |
| "shape": [ |
| 1536 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t094", |
| "name": "encoder.layer.3.intermediate.dense.weight", |
| "shape": [ |
| 1536, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 1536, |
| 384 |
| ] |
| }, |
| { |
| "key": "t095", |
| "name": "encoder.layer.3.layernorm_after.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t096", |
| "name": "encoder.layer.3.layernorm_after.weight", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t097", |
| "name": "encoder.layer.3.layernorm_before.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t098", |
| "name": "encoder.layer.3.layernorm_before.weight", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t099", |
| "name": "encoder.layer.3.output.dense.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t100", |
| "name": "encoder.layer.3.output.dense.weight", |
| "shape": [ |
| 384, |
| 1536 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 1536 |
| ] |
| }, |
| { |
| "key": "t101", |
| "name": "encoder.layer.4.attention.attention.key.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t102", |
| "name": "encoder.layer.4.attention.attention.key.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t103", |
| "name": "encoder.layer.4.attention.attention.query.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t104", |
| "name": "encoder.layer.4.attention.attention.query.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t105", |
| "name": "encoder.layer.4.attention.attention.value.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t106", |
| "name": "encoder.layer.4.attention.attention.value.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t107", |
| "name": "encoder.layer.4.attention.output.dense.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t108", |
| "name": "encoder.layer.4.attention.output.dense.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t109", |
| "name": "encoder.layer.4.intermediate.dense.bias", |
| "shape": [ |
| 1536 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t110", |
| "name": "encoder.layer.4.intermediate.dense.weight", |
| "shape": [ |
| 1536, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 1536, |
| 384 |
| ] |
| }, |
| { |
| "key": "t111", |
| "name": "encoder.layer.4.layernorm_after.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t112", |
| "name": "encoder.layer.4.layernorm_after.weight", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t113", |
| "name": "encoder.layer.4.layernorm_before.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t114", |
| "name": "encoder.layer.4.layernorm_before.weight", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t115", |
| "name": "encoder.layer.4.output.dense.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t116", |
| "name": "encoder.layer.4.output.dense.weight", |
| "shape": [ |
| 384, |
| 1536 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 1536 |
| ] |
| }, |
| { |
| "key": "t117", |
| "name": "encoder.layer.5.attention.attention.key.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t118", |
| "name": "encoder.layer.5.attention.attention.key.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t119", |
| "name": "encoder.layer.5.attention.attention.query.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t120", |
| "name": "encoder.layer.5.attention.attention.query.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t121", |
| "name": "encoder.layer.5.attention.attention.value.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t122", |
| "name": "encoder.layer.5.attention.attention.value.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t123", |
| "name": "encoder.layer.5.attention.output.dense.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t124", |
| "name": "encoder.layer.5.attention.output.dense.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t125", |
| "name": "encoder.layer.5.intermediate.dense.bias", |
| "shape": [ |
| 1536 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t126", |
| "name": "encoder.layer.5.intermediate.dense.weight", |
| "shape": [ |
| 1536, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 1536, |
| 384 |
| ] |
| }, |
| { |
| "key": "t127", |
| "name": "encoder.layer.5.layernorm_after.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t128", |
| "name": "encoder.layer.5.layernorm_after.weight", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t129", |
| "name": "encoder.layer.5.layernorm_before.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t130", |
| "name": "encoder.layer.5.layernorm_before.weight", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t131", |
| "name": "encoder.layer.5.output.dense.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t132", |
| "name": "encoder.layer.5.output.dense.weight", |
| "shape": [ |
| 384, |
| 1536 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 1536 |
| ] |
| }, |
| { |
| "key": "t133", |
| "name": "encoder.layer.6.attention.attention.key.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t134", |
| "name": "encoder.layer.6.attention.attention.key.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t135", |
| "name": "encoder.layer.6.attention.attention.query.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t136", |
| "name": "encoder.layer.6.attention.attention.query.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t137", |
| "name": "encoder.layer.6.attention.attention.value.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t138", |
| "name": "encoder.layer.6.attention.attention.value.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t139", |
| "name": "encoder.layer.6.attention.output.dense.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t140", |
| "name": "encoder.layer.6.attention.output.dense.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t141", |
| "name": "encoder.layer.6.intermediate.dense.bias", |
| "shape": [ |
| 1536 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t142", |
| "name": "encoder.layer.6.intermediate.dense.weight", |
| "shape": [ |
| 1536, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 1536, |
| 384 |
| ] |
| }, |
| { |
| "key": "t143", |
| "name": "encoder.layer.6.layernorm_after.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t144", |
| "name": "encoder.layer.6.layernorm_after.weight", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t145", |
| "name": "encoder.layer.6.layernorm_before.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t146", |
| "name": "encoder.layer.6.layernorm_before.weight", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t147", |
| "name": "encoder.layer.6.output.dense.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t148", |
| "name": "encoder.layer.6.output.dense.weight", |
| "shape": [ |
| 384, |
| 1536 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 1536 |
| ] |
| }, |
| { |
| "key": "t149", |
| "name": "encoder.layer.7.attention.attention.key.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t150", |
| "name": "encoder.layer.7.attention.attention.key.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t151", |
| "name": "encoder.layer.7.attention.attention.query.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t152", |
| "name": "encoder.layer.7.attention.attention.query.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t153", |
| "name": "encoder.layer.7.attention.attention.value.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t154", |
| "name": "encoder.layer.7.attention.attention.value.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t155", |
| "name": "encoder.layer.7.attention.output.dense.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t156", |
| "name": "encoder.layer.7.attention.output.dense.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t157", |
| "name": "encoder.layer.7.intermediate.dense.bias", |
| "shape": [ |
| 1536 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t158", |
| "name": "encoder.layer.7.intermediate.dense.weight", |
| "shape": [ |
| 1536, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 1536, |
| 384 |
| ] |
| }, |
| { |
| "key": "t159", |
| "name": "encoder.layer.7.layernorm_after.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t160", |
| "name": "encoder.layer.7.layernorm_after.weight", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t161", |
| "name": "encoder.layer.7.layernorm_before.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t162", |
| "name": "encoder.layer.7.layernorm_before.weight", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t163", |
| "name": "encoder.layer.7.output.dense.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t164", |
| "name": "encoder.layer.7.output.dense.weight", |
| "shape": [ |
| 384, |
| 1536 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 1536 |
| ] |
| }, |
| { |
| "key": "t165", |
| "name": "encoder.layer.8.attention.attention.key.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t166", |
| "name": "encoder.layer.8.attention.attention.key.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t167", |
| "name": "encoder.layer.8.attention.attention.query.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t168", |
| "name": "encoder.layer.8.attention.attention.query.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t169", |
| "name": "encoder.layer.8.attention.attention.value.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t170", |
| "name": "encoder.layer.8.attention.attention.value.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t171", |
| "name": "encoder.layer.8.attention.output.dense.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t172", |
| "name": "encoder.layer.8.attention.output.dense.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t173", |
| "name": "encoder.layer.8.intermediate.dense.bias", |
| "shape": [ |
| 1536 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t174", |
| "name": "encoder.layer.8.intermediate.dense.weight", |
| "shape": [ |
| 1536, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 1536, |
| 384 |
| ] |
| }, |
| { |
| "key": "t175", |
| "name": "encoder.layer.8.layernorm_after.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t176", |
| "name": "encoder.layer.8.layernorm_after.weight", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t177", |
| "name": "encoder.layer.8.layernorm_before.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t178", |
| "name": "encoder.layer.8.layernorm_before.weight", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t179", |
| "name": "encoder.layer.8.output.dense.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t180", |
| "name": "encoder.layer.8.output.dense.weight", |
| "shape": [ |
| 384, |
| 1536 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 1536 |
| ] |
| }, |
| { |
| "key": "t181", |
| "name": "encoder.layer.9.attention.attention.key.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t182", |
| "name": "encoder.layer.9.attention.attention.key.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t183", |
| "name": "encoder.layer.9.attention.attention.query.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t184", |
| "name": "encoder.layer.9.attention.attention.query.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t185", |
| "name": "encoder.layer.9.attention.attention.value.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t186", |
| "name": "encoder.layer.9.attention.attention.value.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t187", |
| "name": "encoder.layer.9.attention.output.dense.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t188", |
| "name": "encoder.layer.9.attention.output.dense.weight", |
| "shape": [ |
| 384, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 384 |
| ] |
| }, |
| { |
| "key": "t189", |
| "name": "encoder.layer.9.intermediate.dense.bias", |
| "shape": [ |
| 1536 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t190", |
| "name": "encoder.layer.9.intermediate.dense.weight", |
| "shape": [ |
| 1536, |
| 384 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 1536, |
| 384 |
| ] |
| }, |
| { |
| "key": "t191", |
| "name": "encoder.layer.9.layernorm_after.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t192", |
| "name": "encoder.layer.9.layernorm_after.weight", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t193", |
| "name": "encoder.layer.9.layernorm_before.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t194", |
| "name": "encoder.layer.9.layernorm_before.weight", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t195", |
| "name": "encoder.layer.9.output.dense.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t196", |
| "name": "encoder.layer.9.output.dense.weight", |
| "shape": [ |
| 384, |
| 1536 |
| ], |
| "kind": "factor", |
| "rank": 71, |
| "matrix_shape": [ |
| 384, |
| 1536 |
| ] |
| }, |
| { |
| "key": "t197", |
| "name": "layernorm.bias", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| }, |
| { |
| "key": "t198", |
| "name": "layernorm.weight", |
| "shape": [ |
| 384 |
| ], |
| "kind": "replacement" |
| } |
| ], |
| "files_sha256": { |
| "weights.safetensors": "9b58d252978d3a5d33adcdaea95e113162daf5146e20a9e1b8aa47e21e48ed93", |
| "english-assets/README.md": "7e0f183a2697956065aea4e3ec9ca718967be1b7afe446cff4dcdbc8a64fa52b", |
| "english-assets/config.json": "2b6e06edec319984eca99c932d940bfa1f308db3f74d6279440b906c373bf727", |
| "english-assets/generation_config.json": "41149cdcffec4d657f32dfcddd9b208037f01286c9e07945c724908c58ed0193", |
| "english-assets/preprocessor_config.json": "eec0c686d2e2ea6d5887f87e1c6a34ef5bf1ba5b2ac98b8675e131da646fa1d2", |
| "english-assets/sentencepiece.bpe.model": "6f5e2fefcf793761a76a6bfb8ad35489f9c203b25557673284b6d032f41043f4", |
| "english-assets/special_tokens_map.json": "58dc0a3d15e3bad4861bcf2dee770ea327344d185182d4c5d2bf253285e53a83", |
| "english-assets/tokenizer_config.json": "e098c98c6834fbcd3f3e75215e3e8b8868f0fe5bf07aef93d28e4f7196db293c", |
| "character-assets/config.json": "decbe7681cc6000bd5bf83226b295e7a051e8280bc12255961e011d2caadbfce", |
| "character-assets/generation_config.json": "9577d3da76b60ee97a70d1516ac144eba329a8a70dd0e96c6f721c910434dcff", |
| "character-assets/character_codec.json": "7bb65c86bc29b555f8da3af81b7023e370bdebe6ced8c84d104d55165c6ffb0a", |
| "character-assets/preprocessor_config.json": "36a945a7cc645688b9ef64dabae16979cf5f7c1c448569cc306694edc0598b9b", |
| "model-configs.json": "1c598da84fcbe26af738d5ef1c0ea04f80f965e7690b1089aeaa8c7977cd4359", |
| "english-generation.json": "cddf3683b66672c43fee2a6d966829eebf2a00c61d376d0eb30db972693210be", |
| "character-generation.json": "02b8abbca9b3f1e5fb8ea2057b0aea84e1f488dd5dd2ca52a75850170bc3b08b" |
| }, |
| "uniform_rank": 71, |
| "routing_parameters_reserved": 100000, |
| "automatic_dispatch": false, |
| "scope": "Serial CPU source candidate only; no accuracy or native Paddle qualification", |
| "original_model_sha256": "1b83102cbc1520dee1c3937ac334da83da18cf4683d46ebd1bd4e93ebe584dd7", |
| "parent_bundle_sha256": "b9281e7da68d2b9c331697a9fb15da3a2366fb937fce00742d658987d924318e", |
| "training_run_sha256": "713dfd450e19d12523585d9973aad7a7236097e64206e035800bbdcfeacb21e3", |
| "training_step": 8000, |
| "training_chain": "5e79546a138135b739b6485f3af4d81d412baa44119e849ec307ac8d87dff4ec", |
| "initial_character_model_sha256": "fcc1346b8c6aa36ae04b8628deb5cd309aa09d4574f0ed9ee0ce4d124091174a", |
| "root_bundle_sha256": "a52a68c8950c7eada39215d52b7b1eaf893d23577297951019bce382689baf48", |
| "parent_training_step": 8000, |
| "cumulative_training_step": 16000, |
| "parent_training_run_sha256": "f33da3168d15a743a4e73a59a77a55f5c85e41ee675cf006a00c940b9a23c70b" |
| } |
|
|
