Upload model.npz.yml with huggingface_hub
Browse files- model.npz.yml +246 -0
    	
        model.npz.yml
    ADDED
    
    | @@ -0,0 +1,246 @@ | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | 
|  | |
| 1 | 
            +
            authors: false
         | 
| 2 | 
            +
            cite: false
         | 
| 3 | 
            +
            build-info: ""
         | 
| 4 | 
            +
            workspace: -8000
         | 
| 5 | 
            +
            log: train.log
         | 
| 6 | 
            +
            log-level: info
         | 
| 7 | 
            +
            log-time-zone: PST8PDT
         | 
| 8 | 
            +
            quiet: false
         | 
| 9 | 
            +
            quiet-translation: true
         | 
| 10 | 
            +
            seed: 141414
         | 
| 11 | 
            +
            check-nan: false
         | 
| 12 | 
            +
            interpolate-env-vars: true
         | 
| 13 | 
            +
            relative-paths: false
         | 
| 14 | 
            +
            dump-config: ""
         | 
| 15 | 
            +
            sigterm: save-and-exit
         | 
| 16 | 
            +
            model: model_files/model.npz
         | 
| 17 | 
            +
            pretrained-model: ""
         | 
| 18 | 
            +
            ignore-model-config: false
         | 
| 19 | 
            +
            type: transformer
         | 
| 20 | 
            +
            dim-vocabs:
         | 
| 21 | 
            +
              - 64000
         | 
| 22 | 
            +
              - 64000
         | 
| 23 | 
            +
            dim-emb: 1024
         | 
| 24 | 
            +
            factors-dim-emb: 0
         | 
| 25 | 
            +
            factors-combine: sum
         | 
| 26 | 
            +
            lemma-dependency: ""
         | 
| 27 | 
            +
            lemma-dim-emb: 0
         | 
| 28 | 
            +
            dim-rnn: 1024
         | 
| 29 | 
            +
            enc-type: bidirectional
         | 
| 30 | 
            +
            enc-cell: gru
         | 
| 31 | 
            +
            enc-cell-depth: 1
         | 
| 32 | 
            +
            enc-depth: 6
         | 
| 33 | 
            +
            dec-cell: gru
         | 
| 34 | 
            +
            dec-cell-base-depth: 2
         | 
| 35 | 
            +
            dec-cell-high-depth: 1
         | 
| 36 | 
            +
            dec-depth: 6
         | 
| 37 | 
            +
            skip: false
         | 
| 38 | 
            +
            layer-normalization: false
         | 
| 39 | 
            +
            right-left: false
         | 
| 40 | 
            +
            input-types:
         | 
| 41 | 
            +
              []
         | 
| 42 | 
            +
            tied-embeddings: true
         | 
| 43 | 
            +
            tied-embeddings-src: false
         | 
| 44 | 
            +
            tied-embeddings-all: true
         | 
| 45 | 
            +
            output-omit-bias: false
         | 
| 46 | 
            +
            transformer-heads: 8
         | 
| 47 | 
            +
            transformer-no-projection: false
         | 
| 48 | 
            +
            transformer-rnn-projection: false
         | 
| 49 | 
            +
            transformer-pool: false
         | 
| 50 | 
            +
            transformer-dim-ffn: 8192
         | 
| 51 | 
            +
            transformer-decoder-dim-ffn: 8192
         | 
| 52 | 
            +
            transformer-ffn-depth: 2
         | 
| 53 | 
            +
            transformer-decoder-ffn-depth: 0
         | 
| 54 | 
            +
            transformer-ffn-activation: relu
         | 
| 55 | 
            +
            transformer-dim-aan: 2048
         | 
| 56 | 
            +
            transformer-aan-depth: 2
         | 
| 57 | 
            +
            transformer-aan-activation: swish
         | 
| 58 | 
            +
            transformer-aan-nogate: false
         | 
| 59 | 
            +
            transformer-decoder-autoreg: self-attention
         | 
| 60 | 
            +
            transformer-tied-layers: []
         | 
| 61 | 
            +
            transformer-guided-alignment-layer: last
         | 
| 62 | 
            +
            transformer-preprocess: ""
         | 
| 63 | 
            +
            transformer-postprocess-emb: d
         | 
| 64 | 
            +
            transformer-postprocess: dan
         | 
| 65 | 
            +
            transformer-postprocess-top: ""
         | 
| 66 | 
            +
            transformer-train-position-embeddings: false
         | 
| 67 | 
            +
            transformer-depth-scaling: true
         | 
| 68 | 
            +
            transformer-no-bias: false
         | 
| 69 | 
            +
            transformer-no-affine: false
         | 
| 70 | 
            +
            bert-mask-symbol: "[MASK]"
         | 
| 71 | 
            +
            bert-sep-symbol: "[SEP]"
         | 
| 72 | 
            +
            bert-class-symbol: "[CLS]"
         | 
| 73 | 
            +
            bert-masking-fraction: 0.15
         | 
| 74 | 
            +
            bert-train-type-embeddings: true
         | 
| 75 | 
            +
            bert-type-vocab-size: 2
         | 
| 76 | 
            +
            comet-final-sigmoid: false
         | 
| 77 | 
            +
            comet-mix: false
         | 
| 78 | 
            +
            comet-mix-norm: false
         | 
| 79 | 
            +
            comet-dropout: 0.1
         | 
| 80 | 
            +
            comet-mixup: 0
         | 
| 81 | 
            +
            comet-mixup-reg: false
         | 
| 82 | 
            +
            comet-pooler-ffn:
         | 
| 83 | 
            +
              - 2048
         | 
| 84 | 
            +
              - 1024
         | 
| 85 | 
            +
            comet-prepend-zero: false
         | 
| 86 | 
            +
            dropout-rnn: 0
         | 
| 87 | 
            +
            dropout-src: 0
         | 
| 88 | 
            +
            dropout-trg: 0
         | 
| 89 | 
            +
            transformer-dropout: 0.1
         | 
| 90 | 
            +
            transformer-dropout-attention: 0
         | 
| 91 | 
            +
            transformer-dropout-ffn: 0.1
         | 
| 92 | 
            +
            cost-type: ce-sum
         | 
| 93 | 
            +
            multi-loss-type: sum
         | 
| 94 | 
            +
            unlikelihood-loss: false
         | 
| 95 | 
            +
            overwrite: false
         | 
| 96 | 
            +
            overwrite-checkpoint: true
         | 
| 97 | 
            +
            no-reload: false
         | 
| 98 | 
            +
            train-sets:
         | 
| 99 | 
            +
              - stdin
         | 
| 100 | 
            +
            vocabs:
         | 
| 101 | 
            +
              - vocab
         | 
| 102 | 
            +
              - vocab
         | 
| 103 | 
            +
            sentencepiece-alphas:
         | 
| 104 | 
            +
              []
         | 
| 105 | 
            +
            sentencepiece-options: ""
         | 
| 106 | 
            +
            sentencepiece-max-lines: 2000000
         | 
| 107 | 
            +
            no-spm-encode: false
         | 
| 108 | 
            +
            after-epochs: 0
         | 
| 109 | 
            +
            after-batches: 0
         | 
| 110 | 
            +
            after: 40e
         | 
| 111 | 
            +
            disp-freq: 100Mt
         | 
| 112 | 
            +
            disp-first: 10
         | 
| 113 | 
            +
            disp-label-counts: true
         | 
| 114 | 
            +
            save-freq: 1Gt
         | 
| 115 | 
            +
            logical-epoch:
         | 
| 116 | 
            +
              - 1Gt
         | 
| 117 | 
            +
            max-length: 256
         | 
| 118 | 
            +
            max-length-crop: false
         | 
| 119 | 
            +
            tsv: true
         | 
| 120 | 
            +
            tsv-fields: 2
         | 
| 121 | 
            +
            shuffle: batches
         | 
| 122 | 
            +
            no-restore-corpus: true
         | 
| 123 | 
            +
            tempdir: /tmp
         | 
| 124 | 
            +
            sqlite: ""
         | 
| 125 | 
            +
            sqlite-drop: false
         | 
| 126 | 
            +
            devices:
         | 
| 127 | 
            +
              - 0
         | 
| 128 | 
            +
              - 1
         | 
| 129 | 
            +
            no-nccl: false
         | 
| 130 | 
            +
            sharding: local
         | 
| 131 | 
            +
            sync-freq: 200u
         | 
| 132 | 
            +
            cpu-threads: 0
         | 
| 133 | 
            +
            mini-batch: 1000
         | 
| 134 | 
            +
            mini-batch-words: 500000
         | 
| 135 | 
            +
            mini-batch-fit: true
         | 
| 136 | 
            +
            mini-batch-fit-step: 5
         | 
| 137 | 
            +
            gradient-checkpointing: false
         | 
| 138 | 
            +
            maxi-batch: 1000
         | 
| 139 | 
            +
            maxi-batch-sort: trg
         | 
| 140 | 
            +
            shuffle-in-ram: true
         | 
| 141 | 
            +
            data-threads: 8
         | 
| 142 | 
            +
            all-caps-every: 0
         | 
| 143 | 
            +
            english-title-case-every: 0
         | 
| 144 | 
            +
            mini-batch-words-ref: 0
         | 
| 145 | 
            +
            mini-batch-warmup: 4000
         | 
| 146 | 
            +
            mini-batch-track-lr: false
         | 
| 147 | 
            +
            mini-batch-round-up: true
         | 
| 148 | 
            +
            optimizer: adam
         | 
| 149 | 
            +
            optimizer-params:
         | 
| 150 | 
            +
              - 0.9
         | 
| 151 | 
            +
              - 0.999
         | 
| 152 | 
            +
              - 1e-08
         | 
| 153 | 
            +
              - 0.01
         | 
| 154 | 
            +
            optimizer-delay: 1
         | 
| 155 | 
            +
            sync-sgd: true
         | 
| 156 | 
            +
            learn-rate: 0.0005
         | 
| 157 | 
            +
            lr-report: true
         | 
| 158 | 
            +
            lr-decay: 0
         | 
| 159 | 
            +
            lr-decay-strategy: epoch+stalled
         | 
| 160 | 
            +
            lr-decay-start:
         | 
| 161 | 
            +
              - 10
         | 
| 162 | 
            +
              - 1
         | 
| 163 | 
            +
            lr-decay-freq: 50000
         | 
| 164 | 
            +
            lr-decay-reset-optimizer: false
         | 
| 165 | 
            +
            lr-decay-repeat-warmup: false
         | 
| 166 | 
            +
            lr-decay-inv-sqrt:
         | 
| 167 | 
            +
              - 4000
         | 
| 168 | 
            +
            lr-warmup: 4000
         | 
| 169 | 
            +
            lr-warmup-start-rate: 0
         | 
| 170 | 
            +
            lr-warmup-cycle: false
         | 
| 171 | 
            +
            lr-warmup-at-reload: false
         | 
| 172 | 
            +
            label-smoothing: 0.1
         | 
| 173 | 
            +
            factor-weight: 1
         | 
| 174 | 
            +
            clip-norm: 0
         | 
| 175 | 
            +
            exponential-smoothing: 1e-3
         | 
| 176 | 
            +
            exponential-smoothing-replace-freq: 0
         | 
| 177 | 
            +
            guided-alignment: none
         | 
| 178 | 
            +
            guided-alignment-cost: ce
         | 
| 179 | 
            +
            guided-alignment-weight: 0
         | 
| 180 | 
            +
            data-weighting: ""
         | 
| 181 | 
            +
            data-weighting-type: sentence
         | 
| 182 | 
            +
            embedding-vectors:
         | 
| 183 | 
            +
              []
         | 
| 184 | 
            +
            embedding-normalization: false
         | 
| 185 | 
            +
            embedding-fix-src: false
         | 
| 186 | 
            +
            embedding-fix-trg: false
         | 
| 187 | 
            +
            precision:
         | 
| 188 | 
            +
              - float32
         | 
| 189 | 
            +
              - float32
         | 
| 190 | 
            +
            cost-scaling:
         | 
| 191 | 
            +
              - 256.f
         | 
| 192 | 
            +
              - 10000
         | 
| 193 | 
            +
              - 1.f
         | 
| 194 | 
            +
              - 256.f
         | 
| 195 | 
            +
            throw-on-divergence:
         | 
| 196 | 
            +
              []
         | 
| 197 | 
            +
            custom-fallbacks:
         | 
| 198 | 
            +
              []
         | 
| 199 | 
            +
            gradient-norm-average-window: 100
         | 
| 200 | 
            +
            dynamic-gradient-scaling:
         | 
| 201 | 
            +
              - 2
         | 
| 202 | 
            +
              - log
         | 
| 203 | 
            +
            check-gradient-nan: false
         | 
| 204 | 
            +
            normalize-gradient: false
         | 
| 205 | 
            +
            train-embedder-rank:
         | 
| 206 | 
            +
              []
         | 
| 207 | 
            +
            quantize-bits: 0
         | 
| 208 | 
            +
            quantize-optimization-steps: 0
         | 
| 209 | 
            +
            quantize-log-based: false
         | 
| 210 | 
            +
            quantize-biases: false
         | 
| 211 | 
            +
            ulr: false
         | 
| 212 | 
            +
            ulr-query-vectors: ""
         | 
| 213 | 
            +
            ulr-keys-vectors: ""
         | 
| 214 | 
            +
            ulr-trainable-transformation: false
         | 
| 215 | 
            +
            ulr-dim-emb: 0
         | 
| 216 | 
            +
            ulr-dropout: 0
         | 
| 217 | 
            +
            ulr-softmax-temperature: 1
         | 
| 218 | 
            +
            valid-sets:
         | 
| 219 | 
            +
              - dev.en-de
         | 
| 220 | 
            +
            valid-freq: 1Gt
         | 
| 221 | 
            +
            valid-metrics:
         | 
| 222 | 
            +
              - perplexity
         | 
| 223 | 
            +
              - ce-mean-words
         | 
| 224 | 
            +
              - bleu
         | 
| 225 | 
            +
              - chrf
         | 
| 226 | 
            +
            valid-reset-stalled: false
         | 
| 227 | 
            +
            valid-reset-all: false
         | 
| 228 | 
            +
            early-stopping: 40
         | 
| 229 | 
            +
            early-stopping-epsilon:
         | 
| 230 | 
            +
              - 0
         | 
| 231 | 
            +
            early-stopping-on: first
         | 
| 232 | 
            +
            beam-size: 4
         | 
| 233 | 
            +
            normalize: 1.0
         | 
| 234 | 
            +
            max-length-factor: 3
         | 
| 235 | 
            +
            word-penalty: 0.0
         | 
| 236 | 
            +
            allow-unk: false
         | 
| 237 | 
            +
            n-best: false
         | 
| 238 | 
            +
            word-scores: false
         | 
| 239 | 
            +
            valid-mini-batch: 32
         | 
| 240 | 
            +
            valid-max-length: 1000
         | 
| 241 | 
            +
            valid-script-path: ""
         | 
| 242 | 
            +
            valid-script-args:
         | 
| 243 | 
            +
              []
         | 
| 244 | 
            +
            valid-translation-output: valid.trg.output
         | 
| 245 | 
            +
            keep-best: true
         | 
| 246 | 
            +
            valid-log: valid.log
         |