Upload Qwen2VLForConditionalGenerationWithAudio

config.json

@@ -13,11 +13,29 @@
   "audio_pad_token_id": 151658,
   "audio_start_token_id": 151657,
   "bos_token_id": null,
-  "dtype": "bfloat16",
+  "dtype": "float32",
   "eos_token_id": 151645,
   "image_token_id": 151655,
   "model_type": "qwen2_vl",
   "pad_token_id": 151643,
+  "quantization_config": {
+    "_load_in_4bit": true,
+    "_load_in_8bit": false,
+    "bnb_4bit_compute_dtype": "bfloat16",
+    "bnb_4bit_quant_storage": "uint8",
+    "bnb_4bit_quant_type": "nf4",
+    "bnb_4bit_use_double_quant": true,
+    "llm_int8_enable_fp32_cpu_offload": false,
+    "llm_int8_has_fp16_weight": false,
+    "llm_int8_skip_modules": [
+      "audio_encoder",
+      "audio_projector"
+    ],
+    "llm_int8_threshold": 6.0,
+    "load_in_4bit": true,
+    "load_in_8bit": false,
+    "quant_method": "bitsandbytes"
+  },
   "text_config": {
     "attention_dropout": 0.0,
     "bos_token_id": 151643,

generation_config.json

@@ -1,7 +1,7 @@
 {
-  "bos_token_id": 151643,
   "do_sample": true,
   "eos_token_id": [
+    151645,
     151645,
     151643
   ],

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2f1359029f02f8301c962f17ee35cbca59629b1187e17a27246207e9a4684177
-size 19165642192
+oid sha256:787180f929e3ee8c2578857798ea08a7538dc41d2232691fcb4e670aa780edea
+size 8802787595