Training in progress, step 50, checkpoint

checkpoint-50/README.md

@@ -254,4 +254,23 @@ The following `bitsandbytes` quantization config was used during training:
 ### Framework versions
 
 
+- PEFT 0.6.0.dev0
+## 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: 6.0
+- 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
+
+### Framework versions
+
+
 - PEFT 0.6.0.dev0

checkpoint-50/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:52f3e2863a99621fe2385df6cddf1bcd468da4073e58c4120b23dac6c05e517b
+oid sha256:efcd616d78aee5801b26823ff845c7a11cb8437ad33b7f54dd9d8c1d6d55430c
 size 9873829

checkpoint-50/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:583134c5aa4c8a4be35c76d3ef2147918ca197900e94a9bd441bc6b37318017b
+oid sha256:a1c19a314df24e848b955548a7f413702cd6ca621d7171dd290931ec47d2a49f
 size 42724

checkpoint-50/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bc74b07bbeaa07754a3167eed2100e9c261c9fbfeeead9fd2401e092e61ce5ba
+oid sha256:9d5f3946af7bb77e06a589294e88390029883f4d85063278eb29d1e1dd7fdf0d
 size 14244

checkpoint-50/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9cd64bd8bbd74e4d9ac50e02a85e7f6d62a9cf61cc8888584e2b49a46bcf05e3
+oid sha256:7d0fe70547b4c794870547a1163a5866d6ce45ad88bdeba0273a31e4c75090e4
 size 1064

checkpoint-50/trainer_state.json

@@ -1,7 +1,7 @@
 {
-  "best_metric": 0.6568556427955627,
+  "best_metric": 0.665068507194519,
   "best_model_checkpoint": "output/checkpoint-50",
-  "epoch": 0.11347517730496454,
+  "epoch": 0.20460358056265984,
   "eval_steps": 50,
   "global_step": 50,
   "is_hyper_param_search": false,
@@ -10,31 +10,31 @@
   "log_history": [
     {
       "epoch": 0.0,
-      "learning_rate": 1.25e-06,
-      "loss": 0.5043,
+      "learning_rate": 7.575757575757576e-07,
+      "loss": 0.4968,
       "step": 1
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 5.7499999999999995e-05,
-      "loss": 0.4906,
+      "epoch": 0.2,
+      "learning_rate": 3.484848484848485e-05,
+      "loss": 0.5254,
       "step": 50
     },
     {
-      "epoch": 0.11,
-      "eval_accuracy": 0.608955223880597,
-      "eval_loss": 0.6568556427955627,
-      "eval_runtime": 61.5694,
-      "eval_samples_per_second": 5.441,
-      "eval_steps_per_second": 1.364,
+      "epoch": 0.2,
+      "eval_accuracy": 0.6507462686567164,
+      "eval_loss": 0.665068507194519,
+      "eval_runtime": 61.5592,
+      "eval_samples_per_second": 5.442,
+      "eval_steps_per_second": 1.365,
       "step": 50
     }
   ],
   "logging_steps": 50,
-  "max_steps": 440,
-  "num_train_epochs": 1,
+  "max_steps": 732,
+  "num_train_epochs": 3,
   "save_steps": 50,
-  "total_flos": 1.846347914809344e+16,
+  "total_flos": 2.16199352573952e+16,
   "trial_name": null,
   "trial_params": null
 }

checkpoint-50/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:87bd30bd4b21b4a0c66f1fd545858ef6e2d33b5984a8d292e79011acc3affd6b
+oid sha256:9feea516a8a208c393c324d84d047c0ab5036bbdf3fb1f3eea6a2563e9c79a6b
 size 4472