Add new files to repository

config.json

@@ -22,8 +22,9 @@
   "rope_theta": 1000000.0,
   "sliding_window": null,
   "tie_word_embeddings": true,
-  "transformers_version": "4.48.3",
-  "use_cache": false,
+  "torch_dtype": "float32",
+  "transformers_version": "4.49.0",
+  "use_cache": true,
   "use_sliding_window": false,
   "vocab_size": 151936
 }

generation_config.json

@@ -8,5 +8,5 @@
   "pad_token_id": 151643,
   "repetition_penalty": 1.1,
   "top_p": 0.8,
-  "transformers_version": "4.48.3"
+  "transformers_version": "4.49.0"
 }

onnx/model.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:957af0e6e1f0aa66f51413dec161707f7897152e92d602d34e10e6a26158f83f
+size 1856463035

onnx/model_bnb4.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:24b93580201ee1b91e840e70f0fe52dd85dba9ecc401649760982a51883e6038
+size 796771806

onnx/model_fp16.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:665c78c83baba5f31dae00d158b671a36a35a0586eb8564d56003e6ee9d1947a
+size 928498695

onnx/model_int8.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:83e7775ffc274cfce7c1e5565244891bdb598d46e64ccdf4bef497d76f64e0b7
+size 465054956

onnx/model_q4.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eed3fb37e44a804d65449683ceda5648c2211179026f381617d7c17f2c78a37f
+size 816038046

onnx/model_q4f16.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:655dc12eacfeae4bba0dd081ebe1333cd77dd75338d167efd49f3de8f79973df
+size 485368810

onnx/model_quantized.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:83e7775ffc274cfce7c1e5565244891bdb598d46e64ccdf4bef497d76f64e0b7
+size 465054956

onnx/model_uint8.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b16deba4068aed69aece090d9cc8bc26f13d328603fe8f2e2cd2b9f37aa1759d
+size 465055041

quantize_config.json

@@ -0,0 +1,18 @@
+{
+    "modes": [
+        "fp16",
+        "q8",
+        "int8",
+        "uint8",
+        "q4",
+        "q4f16",
+        "bnb4"
+    ],
+    "per_channel": false,
+    "reduce_range": false,
+    "block_size": null,
+    "is_symmetric": true,
+    "accuracy_level": null,
+    "quant_type": 1,
+    "op_block_list": null
+}