Upload MambaForCausalLM

config.json

@@ -1,7 +1,7 @@
 {
   "_name_or_path": "./",
   "architectures": [
-    "FalconMambaForCausalLM"
+    "MambaForCausalLM"
   ],
   "bos_token_id": 0,
   "conv_kernel": 4,
@@ -12,9 +12,9 @@
   "initializer_range": 0.1,
   "intermediate_size": 8192,
   "layer_norm_epsilon": 1e-05,
-  "model_type": "falcon_mamba",
+  "model_type": "mamba",
   "num_hidden_layers": 64,
-  "pad_token_id": 0,
+  "pad_token_id": 11,
   "rescale_prenorm_residual": false,
   "residual_in_fp32": true,
   "state_size": 16,

generation_config.json

@@ -2,6 +2,6 @@
   "_from_model_config": true,
   "bos_token_id": 0,
   "eos_token_id": 11,
-  "pad_token_id": 0,
+  "pad_token_id": 11,
   "transformers_version": "4.43.0.dev0"
 }

model-00001-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7ba523977ce752d3e6681cfad5442934071a748d0c8cd6d7ea3f9843662b0c36
-size 4961951048
+oid sha256:e46c29585a3a779e22609ac6e8e55af8463cb5ff61d3431c36392fc9fcfac523
+size 4956184024

model-00002-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f9201c60ae26961b49c049d390a5c5d302d75261f57e7ea079945bb88d832cdd
-size 4993566096
+oid sha256:e665d3e2f008c56fbe857f510c0dbff539fc196715fe45dc5815f4a12136a471
+size 4987536920

model-00003-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d92d7cef1ff8f6ddf3d3b299360ad92e8d9b3a13a42ed149512d229ebb42c006
-size 4606661504
+oid sha256:0f3bb8f91425be05bfe2781c459d63a81825ece2beb289be5002ef564d7bd492
+size 4601680888

model.safetensors.index.json

@@ -1,6 +1,6 @@
 {
   "metadata": {
-    "total_size": 14562107392
+    "total_size": 14545330176
   },
   "weight_map": {
     "backbone.embeddings.weight": "model-00001-of-00003.safetensors",