Upload checkpoint 2000 (Testing GatedNorm initialization)

config.json

@@ -3,13 +3,17 @@
     "AlinlightForCausalLM"
   ],
   "attention_dropout": 0.0,
+  "attn_logit_softcapping": 50.0,
   "auto_map": {
     "AutoConfig": "configuration_alinlight.AlinlightConfig",
     "AutoModelForCausalLM": "modeling_alinlight.AlinlightForCausalLM"
   },
   "bos_token_id": 1,
   "dtype": "bfloat16",
+  "embed_pdrop": 0.0,
+  "embed_scale": true,
   "eos_token_id": 2,
+  "final_logit_softcapping": 30.0,
   "hidden_size": 2048,
   "initializer_range": 0.02,
   "intermediate_size": 5632,
@@ -19,11 +23,14 @@
   "num_hidden_layers": 22,
   "num_key_value_heads": 8,
   "pad_token_id": 0,
+  "resid_pdrop": 0.0,
   "rms_norm_eps": 1e-05,
   "rope_scaling": null,
   "rope_theta": 10000.0,
   "sliding_window": 4096,
-  "transformers_version": "4.57.6",
+  "transformers_version": "4.57.1",
   "use_cache": false,
-  "vocab_size": 128000
+  "use_qk_norm": true,
+  "vocab_size": 128000,
+  "z_loss_weight": 0.0001
 }

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:495e7b4820b7e7951594664f74f244b9c34ddb5144642fb202ce1da0cdcbcf9c
+size 3033026263

tokenizer.json

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a0839415819a5b37239153cf30aadd092116f0bb1b8e5aa0cc1169c79572f13d
-size 11391355
+oid sha256:d526559046f1e90fa94add973fedc7ccab2b5019d0a69525bd94f6fb448f598f
+size 11391088