Upload folder using huggingface_hub

README.md

@@ -1,4 +1,20 @@
 ---
-license: cc-by-nc-4.0
+library_name: peft
 ---
-This is just a test/checkpoint of qlora on mpt-30b.  You can test if you'd like, but it needs more training to be really usable.
+## Training procedure
+
+
+The following `bitsandbytes` quantization config was used during training:
+- load_in_8bit: False
+- load_in_4bit: True
+- llm_int8_threshold: 6.0
+- llm_int8_skip_modules: None
+- llm_int8_enable_fp32_cpu_offload: False
+- llm_int8_has_fp16_weight: False
+- bnb_4bit_quant_type: nf4
+- bnb_4bit_use_double_quant: True
+- bnb_4bit_compute_dtype: bfloat16
+### Framework versions
+
+
+- PEFT 0.4.0.dev0

adapter_config.json

@@ -13,10 +13,10 @@
   "r": 64,
   "revision": null,
   "target_modules": [
-    "Wqkv",
     "down_proj",
-    "up_proj",
-    "out_proj"
+    "out_proj",
+    "Wqkv",
+    "up_proj"
   ],
   "task_type": "CAUSAL_LM"
 }

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9606c0396f5062dd8aee24a8fe3fded1f48fd463996202703eba23117e5e69a7
+oid sha256:e82b2d0efd1143acd0b12bbcbc45d0b3951b24ea7011d50008e4b8aa0b50ad3b
 size 704782669