wnut-no-crf.json

{
    "task": "tagger",
    "basedir": "wnut-multi-emb-iobes",
    "batchsz": 20,
    "conll_output": "wnutresults.conll",
    "unif": 0.1,
    "preproc": {
	"mxlen": 60,
	"mxwlen": 40
    },
    "features": [
	{
	    "name": "word",
	    "vectorizer": {
		"type": "dict1d",
		"fields": "text",
                "transform": "baseline.lowercase"
	    },
	    "embeddings": { "label": "glove-twitter-27B"}
	},
	{
	    "name": "word2",
	    "vectorizer": {
		"type": "dict1d",
		"fields": "text",
                "transform": "baseline.lowercase"
	    },
	    "embeddings": { "label": "glove-42B"}
	},
        {
            "name": "word3",
            "vectorizer": {
                "type": "dict1d",
                "fields": "text",
                "transform": "baseline.web_cleanup"
            },
            "embeddings": { "label": "w2v-twitter-30M" }
        },
	{
	    "name": "char",
	    "vectorizer": { "type": "dict2d" },
	    "embeddings": { "dsz": 30, "wsz": 30, "type": "char-conv" }
	}
    ],
    "backend": "pytorch",
    "dataset": "wnut-iobes",
    "loader": {
	"reader_type": "default",
        "named_fields": {
          "0":"text",
          "-1":"y"
        }
    },
    "model": {
	"model_type": "default",
	"cfiltsz": [
	    3
	],
	"hsz": 200,
	"dropout": 0.5,
	"rnntype": "blstm",
	"layers": 1,
	"constrain_decode":true,
	"crf": 0 
    },
    "train": {
	"epochs": 60,
	"optim": "sgd",
	"decay": 0,
	"eta": 0.015,
	"mom": 0.9,
	"patience": 20,
	"early_stopping_metric": "f1",
	"clip": 5.0,
        "span_type": "iobes"
    }
}