Upload 9 files

config.json

@@ -1,8 +1,9 @@
 {
-  "_name_or_path": "/storage0/bi/models/FRED-T5-large-instruct-v2",
+  "_name_or_path": "/storage0/bi/models/FRED-T5-large-rag",
   "architectures": [
     "T5ForConditionalGeneration"
   ],
+  "attention_type": "flash",
   "bos_token_id": 1,
   "classifier_dropout": 0.0,
   "d_ff": 2816,
@@ -13,6 +14,7 @@
   "dropout_rate": 0.1,
   "eos_token_id": 2,
   "feed_forward_proj": "gated-gelu",
+  "global_block_size": 32,
   "gradient_checkpointing": false,
   "initializer_factor": 1.0,
   "is_encoder_decoder": true,
@@ -26,9 +28,11 @@
   "pad_token_id": 0,
   "relative_attention_max_distance": 128,
   "relative_attention_num_buckets": 32,
+  "sliding_window": 512,
   "tie_word_embeddings": false,
   "torch_dtype": "float32",
   "transformers_version": "4.40.2",
   "use_cache": true,
+  "use_triton": true,
   "vocab_size": 50364
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c02c30ef95376872eae96db748c866b5e95e681e3b2de63449cef97e673db251
+oid sha256:48699c59631d9d38cd7672bb6c68d46122a7c0bbd23cefe426353294f9befbfa
 size 3282058120

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e93971f1421a8bbb3f14d268992073d1f1a6ab64f75815e201aae6b722e20531
+oid sha256:c53ec08254cd0a70d22f7ddf0c1c8003af301c6a6fee5683042178c730789e26
 size 3282175109