Training loss: 6.7703

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "nroggendorff/smallama-it",
+  "_name_or_path": "nroggendorff/smallama",
   "architectures": [
     "LlamaForCausalLM"
   ],
@@ -7,17 +7,17 @@
   "attention_dropout": 0.0,
   "bos_token_id": 0,
   "eos_token_id": 2,
-  "head_dim": 48,
+  "head_dim": 16,
   "hidden_act": "silu",
-  "hidden_size": 576,
+  "hidden_size": 864,
   "initializer_range": 0.02,
-  "intermediate_size": 2304,
-  "max_position_embeddings": 128,
+  "intermediate_size": 3456,
+  "max_position_embeddings": 512,
   "mlp_bias": false,
   "model_type": "llama",
-  "num_attention_heads": 12,
-  "num_hidden_layers": 12,
-  "num_key_value_heads": 12,
+  "num_attention_heads": 54,
+  "num_hidden_layers": 27,
+  "num_key_value_heads": 54,
   "pad_token_id": 1,
   "pretraining_tp": 1,
   "rms_norm_eps": 1e-05,
@@ -27,5 +27,5 @@
   "torch_dtype": "float32",
   "transformers_version": "4.46.3",
   "use_cache": true,
-  "vocab_size": 32003
+  "vocab_size": 1111
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e2ba43a9f491d427d32b4da4708d0f53a6fd30e118be53475225fb68cbe4abc2
-size 402343712
+oid sha256:cc186bbad0258166501977537701ce28a7a10d8d300a735b0eadab2033af519a
+size 1297841912