Update config.json

config.json

@@ -32,9 +32,25 @@
   "pad_token_id": 128001,
   "mlp_bias": false,
   "head_dim": 128,
-  "quantization_config": {
-    "load_in_8bit": false,
-    "load_in_4bit": false
+  "attention_implementation": "flash_attention_2",
+  "use_sliding_window": false,
+  "sliding_window": null,
+  "quantization_config": null,
+  "safetensors": {
+    "enabled": true,
+    "total_shards": 96,
+    "shard_pattern": "model-{:05d}-of-00096.safetensors",
+    "index_file": "model.safetensors.index.json",
+    "fast_loading": true,
+    "zero_copy": true
+  },
+  "optimization": {
+    "flash_attention": true,
+    "gradient_checkpointing": false,
+    "tensor_parallel_size": 2,
+    "pipeline_parallel_size": 1,
+    "gpu_memory_fraction": 0.95,
+    "max_batch_size": 32
   },
   "auto_map": {
     "AutoConfig": "configuration_llama.LlamaConfig",