Training in progress, epoch 1, checkpoint

checkpoint-4/README.md

@@ -4,6 +4,18 @@ library_name: peft
 ## Training procedure
 
 
+The following `bitsandbytes` quantization config was used during training:
+- quant_method: bitsandbytes
+- load_in_8bit: True
+- load_in_4bit: False
+- llm_int8_threshold: 4.0
+- llm_int8_skip_modules: None
+- llm_int8_enable_fp32_cpu_offload: False
+- llm_int8_has_fp16_weight: False
+- bnb_4bit_quant_type: fp4
+- bnb_4bit_use_double_quant: False
+- bnb_4bit_compute_dtype: float32
+
 The following `bitsandbytes` quantization config was used during training:
 - quant_method: bitsandbytes
 - load_in_8bit: True
@@ -17,5 +29,6 @@ The following `bitsandbytes` quantization config was used during training:
 - bnb_4bit_compute_dtype: bfloat16
 ### Framework versions
 
+- PEFT 0.5.0
 
 - PEFT 0.5.0

checkpoint-4/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:69d01cdfa8b30ff745d5e612123a815e0dbf4ddb3e1700172d7314d8d02e5d09
-size 30440349
+oid sha256:799a00a1ef0e9c6047cdfcb73b2c05593ee5344c13c74ceebfe1f44b4fafff9c
+size 30047133

checkpoint-4/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7815b4c9010c60064ea3325b4eee9b32c34b150449f33c6c62e8951c223fbe3d
-size 60898982
+oid sha256:6184825974fd529dffbb25d8772c9e0dc8a28cd66a92f5e2bfb7a47c9a85ac9d
+size 60113190

checkpoint-4/trainer_state.json

@@ -11,9 +11,9 @@
     {
       "epoch": 1.0,
       "eval_loss": NaN,
-      "eval_runtime": 11.7568,
-      "eval_samples_per_second": 430.984,
-      "eval_steps_per_second": 13.524,
+      "eval_runtime": 9.3163,
+      "eval_samples_per_second": 543.884,
+      "eval_steps_per_second": 17.067,
       "step": 4
     }
   ],

checkpoint-4/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:238151b110e2e4dfc6ebf1dd217dc2b2ee63a49132a41b9b3baa63652e9edd25
+oid sha256:c851a3ebc58e1c8c186392bf892cbae44ea2e2e184458efd3713128db6d5e05d
 size 4219