config.json (#2)
Browse files- config.json (cff9337d2eab1afc3ad1c8f0f09e01d0a21e7882)
- config.json +70 -0
config.json
ADDED
|
@@ -0,0 +1,70 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"SwitchTransformersForConditionalGeneration"
|
| 4 |
+
],
|
| 5 |
+
"batch_prioritized_routing": false,
|
| 6 |
+
"d_ff": 10240,
|
| 7 |
+
"d_kv": 64,
|
| 8 |
+
"d_model": 4096,
|
| 9 |
+
"decoder_sparse_step": 2,
|
| 10 |
+
"dense_act_fn": "gelu",
|
| 11 |
+
"dropout_rate": 0.1,
|
| 12 |
+
"encoder_sparse_step": 2,
|
| 13 |
+
"eos_token_id": 1,
|
| 14 |
+
"expert_capacity": 64,
|
| 15 |
+
"feed_forward_proj": "gelu",
|
| 16 |
+
"initializer_factor": 1.0,
|
| 17 |
+
"decoder_start_token_id": 0,
|
| 18 |
+
"is_encoder_decoder": true,
|
| 19 |
+
"is_gated_act": false,
|
| 20 |
+
"layer_norm_epsilon": 1e-06,
|
| 21 |
+
"model_type": "switch_transformers",
|
| 22 |
+
"num_decoder_layers": 12,
|
| 23 |
+
"num_experts": 128,
|
| 24 |
+
"num_heads": 64,
|
| 25 |
+
"num_layers": 48,
|
| 26 |
+
"num_selected_experts": 1,
|
| 27 |
+
"num_sparse_decoder_layers": 12,
|
| 28 |
+
"num_sparse_encoder_layers": 12,
|
| 29 |
+
"pad_token_id": 0,
|
| 30 |
+
"relative_attention_max_distance": 128,
|
| 31 |
+
"relative_attention_num_buckets": 32,
|
| 32 |
+
"router_bias": false,
|
| 33 |
+
"router_dtype": "float32",
|
| 34 |
+
"router_ignore_padding_tokens": false,
|
| 35 |
+
"router_jitter_noise": 0,
|
| 36 |
+
"router_type": "tokens_masked",
|
| 37 |
+
"task_specific_params": {
|
| 38 |
+
"summarization": {
|
| 39 |
+
"early_stopping": true,
|
| 40 |
+
"length_penalty": 2.0,
|
| 41 |
+
"max_length": 200,
|
| 42 |
+
"min_length": 30,
|
| 43 |
+
"no_repeat_ngram_size": 3,
|
| 44 |
+
"num_beams": 4,
|
| 45 |
+
"prefix": "summarize: "
|
| 46 |
+
},
|
| 47 |
+
"translation_en_to_de": {
|
| 48 |
+
"early_stopping": true,
|
| 49 |
+
"max_length": 300,
|
| 50 |
+
"num_beams": 4,
|
| 51 |
+
"prefix": "translate English to German: "
|
| 52 |
+
},
|
| 53 |
+
"translation_en_to_fr": {
|
| 54 |
+
"early_stopping": true,
|
| 55 |
+
"max_length": 300,
|
| 56 |
+
"num_beams": 4,
|
| 57 |
+
"prefix": "translate English to French: "
|
| 58 |
+
},
|
| 59 |
+
"translation_en_to_ro": {
|
| 60 |
+
"early_stopping": true,
|
| 61 |
+
"max_length": 300,
|
| 62 |
+
"num_beams": 4,
|
| 63 |
+
"prefix": "translate English to Romanian: "
|
| 64 |
+
}
|
| 65 |
+
},
|
| 66 |
+
"torch_dtype": "bfloat16",
|
| 67 |
+
"transformers_version": "4.24.0.dev0",
|
| 68 |
+
"use_cache": true,
|
| 69 |
+
"vocab_size": 32128
|
| 70 |
+
}
|