Upload tiny models for SwitchTransformersModel

config.json

@@ -7,20 +7,19 @@
   "d_ff": 37,
   "d_kv": 8,
   "d_model": 32,
-  "decoder_sparse_step": 1,
-  "decoder_start_token_id": 0,
+  "decoder_sparse_step": 4,
   "dense_act_fn": "relu",
   "dropout_rate": 0.1,
   "encoder_sparse_step": 1,
   "eos_token_id": 1,
-  "expert_capacity": 100,
+  "expert_capacity": 64,
   "feed_forward_proj": "relu",
   "initializer_factor": 0.002,
-  "is_encoder_decoder": true,
+  "is_encoder_decoder": false,
   "is_gated_act": false,
   "layer_norm_epsilon": 1e-06,
   "model_type": "switch_transformers",
-  "num_decoder_layers": 5,
+  "num_decoder_layers": 12,
   "num_experts": 8,
   "num_heads": 4,
   "num_layers": 5,
@@ -33,11 +32,11 @@
   "router_bias": false,
   "router_dtype": "float32",
   "router_ignore_padding_tokens": false,
-  "router_jitter_noise": 0.0,
+  "router_jitter_noise": 0.01,
   "router_type": "tokens_masked",
   "router_z_loss_coef": 0.001,
   "torch_dtype": "float32",
-  "transformers_version": "4.25.0.dev0",
+  "transformers_version": "4.28.0.dev0",
   "use_cache": true,
-  "vocab_size": 1302
+  "vocab_size": 32100
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0051fce65e67844ca03abfedc1fdef3edee1c284d59d146351dc4f8099666d1f
-size 546373
+oid sha256:a36a5a9e45ffeaed52793d7dbc92697056130f57630bf86a3a19c02933f10aab
+size 5033829

spiece.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d60acb128cf7b7f2536e8f38a5b18a05535c9e14c7a355904270e15b0945ea86
+size 791656

tokenizer_config.json

@@ -104,8 +104,8 @@
   "eos_token": "</s>",
   "extra_ids": 100,
   "model_max_length": 512,
-  "name_or_path": "google/switch-base-8",
   "pad_token": "<pad>",
+  "sp_model_kwargs": {},
   "special_tokens_map_file": null,
   "tokenizer_class": "T5Tokenizer",
   "unk_token": "<unk>"