Upload FlashGPTNeoXForCausalLM

config.json

@@ -1,7 +1,7 @@
 {
-  "_name_or_path": "cleaner_model_v3.pth",
+  "_name_or_path": "EleutherAI/pythia-1.4b-deduped",
   "architectures": [
-    "GPTNeoXForCausalLM"
+    "FlashGPTNeoXForCausalLM"
   ],
   "attention_dropout": 0.0,
   "bos_token_id": 0,
@@ -15,8 +15,8 @@
   "layer_norm_eps": 1e-05,
   "max_position_embeddings": 2048,
   "model_type": "gpt_neox",
-  "num_attention_heads": 8,
-  "num_hidden_layers": 16,
+  "num_attention_heads": 16,
+  "num_hidden_layers": 24,
   "pad_token_id": 1,
   "rope_scaling": {
     "factor": 4.0,

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a75b5234fd44205510c2e89514fb943c936ca1a532fa443078ea362f6d635fa5
-size 4047197149
+oid sha256:b4cff9f2a3a6a3e04bc58f56ebdbf39e2ac88ec28f7c4e8aa19b711b71bc0b11
+size 5658700617