Upload model state (Full) via LLM Platform

Saved from LLM Platform UI. Model class: LlamaForCausalLM. State: Full Model.

config.json

@@ -10,8 +10,8 @@
   ],
   "attention_bias": false,
   "attention_dropout": 0.0,
-  "attention_variant": "flash_attention_2",
-  "attn_implementation": "flash_attention_2",
+  "attention_variant": "auto",
+  "attn_implementation": "auto",
   "auto_optimization_enabled": false,
   "baseline_distribution": null,
   "bitnet_applied": false,
@@ -89,7 +89,7 @@
   "num_hidden_layers": 1,
   "num_key_value_heads": 1,
   "off_topic_filter": false,
-  "optimizer": "adamw",
+  "optimizer": "adamw_torch",
   "original_num_layers": 1,
   "pad_token_id": 2,
   "peft_adapter_added": false,
@@ -102,7 +102,7 @@
   "profanity_filter": false,
   "pruning_amount": null,
   "pruning_applied": false,
-  "qa_restrictions_removed": false,
+  "qa_restrictions_removed": true,
   "qat_applied": false,
   "quantization_applied": false,
   "quantization_mode": "float32",
@@ -120,6 +120,7 @@
   "rms_norm_eps": 1e-05,
   "rope_scaling": {
     "factor": 2.0,
+    "rope_type": "linear",
     "type": "linear"
   },
   "rope_scaling_factor": 2.0,
@@ -144,9 +145,9 @@
   "toxicity_filter": false,
   "transformers_version": "4.50.0.dev0",
   "ultra_fast_mode": false,
-  "untied_embeddings": false,
+  "untied_embeddings": true,
   "use_cache": false,
-  "use_flash_attention_2": true,
+  "use_flash_attention_2": false,
   "vocab_size": 32002,
   "weight_decay_disabled": false,
   "weight_init_applied": false

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9637659163cf84d104e98357e29e54677569a502a6ec22688980701374af790b
-size 688936172
+oid sha256:9aaaa407759ba85a4710ddadec26cb41ccae754db050b26726ca444378a256d0
+size 51960328