Update config.json

config.json

@@ -36,6 +36,15 @@
   "q_lora_rank": 1536,
   "qk_nope_head_dim": 128,
   "qk_rope_head_dim": 64,
+  "quantization_config": {
+    "activation_scheme": "dynamic",
+    "fmt": "e4m3",
+    "quant_method": "fp8",
+    "weight_block_size": [
+      128,
+      128
+    ]
+  },
   "rms_norm_eps": 1e-06,
   "rope_scaling": {
     "beta_fast": 32,
@@ -54,7 +63,7 @@
   "topk_group": 4,
   "topk_method": "noaux_tc",
   "torch_dtype": "bfloat16",
-  "transformers_version": "4.46.3",
+  "transformers_version": "4.33.1",
   "use_cache": true,
   "v_head_dim": 128,
   "vocab_size": 129280