Upload int8 quantized model (manual quantize)

config.json

@@ -5,7 +5,7 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "bos_token_id": 32013,
-  "dtype": "bfloat16",
+  "dtype": "float16",
   "eos_token_id": 32021,
   "head_dim": 128,
   "hidden_act": "silu",

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5b5f94f3913b4b8fc7ef2e032dec2e37d98f52f8d08b66abd04f578e0760f0e0
-size 6887114471
+oid sha256:36aa097f79ff14fce3d0fd7dce2512e9cab8bcdb4bf1016733e4cf9c74cd27e6
+size 6740611183