2023-08-04 18:53:55 Autosave for checkpoint additions

README.md

@@ -37,6 +37,17 @@ The following `bitsandbytes` quantization config was used during training:
 - bnb_4bit_use_double_quant: True
 - bnb_4bit_compute_dtype: bfloat16
 
+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
+
 The following `bitsandbytes` quantization config was used during training:
 - load_in_8bit: False
 - load_in_4bit: True
@@ -52,5 +63,6 @@ The following `bitsandbytes` quantization config was used during training:
 - PEFT 0.5.0.dev0
 - PEFT 0.5.0.dev0
 - PEFT 0.5.0.dev0
+- PEFT 0.5.0.dev0
 
 - PEFT 0.5.0.dev0

adapter_config.json

@@ -3,7 +3,7 @@
   "base_model_name_or_path": "/workspace/webui/models/TheBloke_Llama-2-13B-fp16",
   "bias": "none",
   "fan_in_fan_out": null,
-  "inference_mode": true,
+  "inference_mode": false,
   "init_lora_weights": true,
   "layers_pattern": null,
   "layers_to_transform": null,
@@ -14,13 +14,13 @@
   "r": 32,
   "revision": null,
   "target_modules": [
-    "k_proj",
-    "q_proj",
     "v_proj",
     "o_proj",
     "gate_proj",
+    "k_proj",
+    "up_proj",
     "down_proj",
-    "up_proj"
+    "q_proj"
   ],
   "task_type": "CAUSAL_LM"
 }

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:43d8489e1860bc44ac0a168b429d6db410e83fb55df06da817c1b03bc8a784c7
+oid sha256:bb27235038e583f4bb42db82005cef114ed1580c627dd215688b950029824795
 size 500897101