Auto upload zain 2026-08-14T21:14:29.212142

zain/Activation/out/glu-gelu-100L_trash_run/checkpoint-2700/config.json

@@ -0,0 +1,36 @@
+{
+  "activation": "gelu",
+  "architectures": [
+    "TinyLlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 1,
+  "dtype": "bfloat16",
+  "eos_token_id": 2,
+  "head_dim": 32,
+  "hidden_act": "silu",
+  "hidden_size": 128,
+  "initializer_range": 0.02,
+  "intermediate_size": 256,
+  "max_position_embeddings": 512,
+  "mlp_bias": false,
+  "mlp_type": "glu",
+  "model_type": "tiny_llama",
+  "num_attention_heads": 4,
+  "num_hidden_layers": 100,
+  "num_key_value_heads": 4,
+  "pad_token_id": 0,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 10000.0,
+    "rope_type": "default"
+  },
+  "tie_word_embeddings": true,
+  "tokenizer_name": "w-ahmad/tiny-stories-tokenizer",
+  "transformers_version": "5.16.0.dev0",
+  "use_cache": false,
+  "vocab_size": 4096,
+  "waleed_beta": 4
+}

zain/Activation/out/glu-gelu-100L_trash_run/checkpoint-2700/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3e07ca7de94736a00606ce36dddc7587c628742d4ec9cf2ef5eb18b1517ab07d
+size 33967272

zain/Activation/out/glu-gelu-100L_trash_run/checkpoint-2700/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:af1d5e3bdc382f90bd2bc2b3079fb91901a280e0894bc516b9bd9295cdbf7d81
+size 39573050

zain/Activation/out/glu-gelu-100L_trash_run/checkpoint-2700/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3cf9097d4513154245c48236b6ec5137b7ee2a21c9f58f2cba798ea275c6026f
+size 14244

zain/Activation/out/glu-gelu-100L_trash_run/checkpoint-2700/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6734940bf02a09ae82e3bc3a7a4079331a195a053fcffc27fad5719bee088bc4
+size 1064