End of training

README.md

@@ -1,9 +1,11 @@
 ---
 base_model: sohyunan/gemma-2-2b-it_controller_sft_random
+datasets: maze_5x5
 library_name: transformers
 model_name: debug
 tags:
 - generated_from_trainer
+- controller-grpo
 - trl
 - grpo
 licence: license
@@ -11,7 +13,7 @@ licence: license
 
 # Model Card for debug
 
-This model is a fine-tuned version of [sohyunan/gemma-2-2b-it_controller_sft_random](https://huggingface.co/sohyunan/gemma-2-2b-it_controller_sft_random).
+This model is a fine-tuned version of [sohyunan/gemma-2-2b-it_controller_sft_random](https://huggingface.co/sohyunan/gemma-2-2b-it_controller_sft_random) on the [maze_5x5](https://huggingface.co/datasets/maze_5x5) dataset.
 It has been trained using [TRL](https://github.com/huggingface/trl).
 
 ## Quick start

config.json

@@ -0,0 +1,37 @@
+{
+  "_attn_implementation_autoset": true,
+  "_name_or_path": "sohyunan/gemma-2-2b-it_controller_sft_random",
+  "architectures": [
+    "Gemma2ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_logit_softcapping": 50.0,
+  "bos_token_id": 2,
+  "cache_implementation": "hybrid",
+  "eos_token_id": [
+    1,
+    107
+  ],
+  "final_logit_softcapping": 30.0,
+  "head_dim": 256,
+  "hidden_act": "gelu_pytorch_tanh",
+  "hidden_activation": "gelu_pytorch_tanh",
+  "hidden_size": 2304,
+  "initializer_range": 0.02,
+  "intermediate_size": 9216,
+  "max_position_embeddings": 8192,
+  "model_type": "gemma2",
+  "num_attention_heads": 8,
+  "num_hidden_layers": 26,
+  "num_key_value_heads": 4,
+  "pad_token_id": 0,
+  "query_pre_attn_scalar": 256,
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 10000.0,
+  "sliding_window": 4096,
+  "torch_dtype": "float32",
+  "transformers_version": "4.49.0.dev0",
+  "use_cache": true,
+  "vocab_size": 256001
+}