add model

config.json

@@ -0,0 +1,34 @@
+{
+  "_name_or_path": "rankgen-models-hf/rankgen-t5-xl-all",
+  "architectures": [
+    "T5EncoderWithProjection"
+  ],
+  "auto_map": {
+    "AutoModel": "modeling_rankgen.T5EncoderWithProjection"
+  },
+  "d_ff": 5120,
+  "d_kv": 64,
+  "d_model": 2048,
+  "decoder_start_token_id": 0,
+  "dense_act_fn": "gelu_new",
+  "dropout_rate": 0.1,
+  "eos_token_id": 1,
+  "feed_forward_proj": "gated-gelu",
+  "initializer_factor": 1.0,
+  "is_encoder_decoder": true,
+  "is_gated_act": true,
+  "layer_norm_epsilon": 1e-06,
+  "model_type": "t5",
+  "num_decoder_layers": 24,
+  "num_heads": 32,
+  "num_layers": 24,
+  "output_past": true,
+  "pad_token_id": 0,
+  "relative_attention_max_distance": 128,
+  "relative_attention_num_buckets": 32,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float32",
+  "transformers_version": "4.20.1",
+  "use_cache": true,
+  "vocab_size": 32128
+}

modeling_rankgen.py

@@ -0,0 +1,19 @@
+import torch
+import tqdm
+from torch import nn
+from transformers import T5PreTrainedModel, T5EncoderModel
+
+class T5EncoderWithProjection(T5PreTrainedModel):
+    def __init__(self, config):
+        super().__init__(config)
+        self.config = config
+        self.t5_encoder = T5EncoderModel(config)
+        self.projection = nn.Linear(config.d_model, config.d_model, bias=False)
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def forward(self, **input_args):
+        hidden_states = self.t5_encoder(**input_args).last_hidden_state
+        hidden_states = hidden_states[:, 0, :]
+        batch_embeddings = self.projection(hidden_states)
+        return batch_embeddings

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:846444371a6be0d9d58ff0c4bffff53e460881a5b28e0a513d4460b92d88d784
+size 4910960649