Upload STLForCausalLM

README.md

@@ -1,4 +1,4 @@
 ---
 datasets:
 - saracandu/stl_formulae
----
+---

config.json

@@ -1,6 +1,10 @@
 {
+  "_name_or_path": "saracandu/stldec_random",
   "activation_dropout": 0.0,
   "activation_function": "gelu",
+  "architectures": [
+    "STLForCausalLM"
+  ],
   "attention_dropout": 0.0,
   "bos_token_id": 2,
   "d_model": 1024,
@@ -18,14 +22,16 @@
   "eos_token_id": 3,
   "forced_eos_token_id": 3,
   "init_std": 0.02,
-  "is_encoder_decoder": true,
+  "is_decoder": true,
+  "is_encoder_decoder": false,
   "max_position_embeddings": 1024,
   "model_type": "stldec",
   "num_hidden_layers": 12,
   "pad_token_id": 1,
   "scale_embedding": false,
   "share_encoder_decoder_embeddings": true,
-  "transformers_version": "4.51.3",
+  "torch_dtype": "float32",
+  "transformers_version": "4.47.1",
   "use_cache": true,
   "vocab_size": 35
 }

generation_config.json

@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 2,
+  "decoder_start_token_id": 3,
+  "eos_token_id": 3,
+  "forced_eos_token_id": 3,
+  "pad_token_id": 1,
+  "transformers_version": "4.47.1"
+}