Training in progress, step 200, checkpoint

checkpoint-200/README.md

@@ -216,4 +216,23 @@ 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-200/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7269b850f6f41ba7c1d66ea749e7522072b975c8b7566fffb59bf81b0ec49e7e
+oid sha256:02fc7ae11f75cf918dede926679dd7d282377662923fe8f232e583ac7b5ffcaa
 size 9873829

checkpoint-200/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:031e3b86982e5c64ef004e11b13cb4f773afd6611c65834f21e2fcfbd45b2bab
+oid sha256:44c61db8c54f88280856241d25545a2f77f192b2abacb4fd5466a48cb5b0cfb3
 size 42724

checkpoint-200/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8f467c31b4b77b1fb94b7449189df5c441d6ed0a8c79be61c95999fc9950eb31
+oid sha256:87284756e2bb4c6eff667a34242b85381924820ec6fe0fe5fd799db8c3103821
 size 14244

checkpoint-200/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:74ab2cfe8bb88e0e1613adb80a019b7fd597126800ba6c4a209650193e26a7d9
+oid sha256:653c5d079a71d092b2d513b1b568a861c98ec213041332341725cab9e70ed0ed
 size 1064

checkpoint-200/trainer_state.json

@@ -1,7 +1,7 @@
 {
-  "best_metric": 0.7115468382835388,
+  "best_metric": 0.6528732180595398,
   "best_model_checkpoint": "output/checkpoint-200",
-  "epoch": 0.20052638175209925,
+  "epoch": 0.16043316955780607,
   "eval_steps": 50,
   "global_step": 200,
   "is_hyper_param_search": false,
@@ -10,76 +10,76 @@
   "log_history": [
     {
       "epoch": 0.0,
-      "learning_rate": 1.6666666666666667e-06,
-      "loss": 1.2477,
+      "learning_rate": 1.3333333333333332e-06,
+      "loss": 0.785,
       "step": 1
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 7.5e-05,
-      "loss": 0.9677,
+      "epoch": 0.04,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 0.6547,
       "step": 50
     },
     {
-      "epoch": 0.05,
-      "eval_accuracy": 0.6925373134328359,
-      "eval_loss": 0.6988171339035034,
-      "eval_runtime": 62.4321,
-      "eval_samples_per_second": 5.366,
-      "eval_steps_per_second": 1.073,
+      "epoch": 0.04,
+      "eval_accuracy": 0.6626865671641791,
+      "eval_loss": 0.8300915956497192,
+      "eval_runtime": 61.5689,
+      "eval_samples_per_second": 5.441,
+      "eval_steps_per_second": 1.364,
       "step": 50
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 0.00015833333333333332,
-      "loss": 0.7171,
+      "epoch": 0.08,
+      "learning_rate": 0.00012666666666666666,
+      "loss": 0.6654,
       "step": 100
     },
     {
-      "epoch": 0.1,
-      "eval_accuracy": 0.6805970149253732,
-      "eval_loss": 0.7650710344314575,
-      "eval_runtime": 62.3301,
-      "eval_samples_per_second": 5.375,
-      "eval_steps_per_second": 1.075,
+      "epoch": 0.08,
+      "eval_accuracy": 0.6895522388059702,
+      "eval_loss": 0.6663276553153992,
+      "eval_runtime": 56.4633,
+      "eval_samples_per_second": 5.933,
+      "eval_steps_per_second": 1.488,
       "step": 100
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 0.00024166666666666664,
-      "loss": 0.6936,
+      "epoch": 0.12,
+      "learning_rate": 0.00019333333333333333,
+      "loss": 0.6491,
       "step": 150
     },
     {
-      "epoch": 0.15,
-      "eval_accuracy": 0.6358208955223881,
-      "eval_loss": 0.9820537567138672,
-      "eval_runtime": 62.565,
-      "eval_samples_per_second": 5.354,
-      "eval_steps_per_second": 1.071,
+      "epoch": 0.12,
+      "eval_accuracy": 0.6537313432835821,
+      "eval_loss": 0.9172552824020386,
+      "eval_runtime": 56.4705,
+      "eval_samples_per_second": 5.932,
+      "eval_steps_per_second": 1.488,
       "step": 150
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 0.00029449204406364746,
-      "loss": 0.6546,
+      "epoch": 0.16,
+      "learning_rate": 0.00026,
+      "loss": 0.641,
       "step": 200
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.16,
       "eval_accuracy": 0.6865671641791045,
-      "eval_loss": 0.7115468382835388,
-      "eval_runtime": 62.8545,
-      "eval_samples_per_second": 5.33,
-      "eval_steps_per_second": 1.066,
+      "eval_loss": 0.6528732180595398,
+      "eval_runtime": 56.4485,
+      "eval_samples_per_second": 5.935,
+      "eval_steps_per_second": 1.488,
       "step": 200
     }
   ],
   "logging_steps": 50,
-  "max_steps": 997,
+  "max_steps": 1246,
   "num_train_epochs": 1,
   "save_steps": 100,
-  "total_flos": 9.64590287374848e+16,
+  "total_flos": 7.505235259858944e+16,
   "trial_name": null,
   "trial_params": null
 }

checkpoint-200/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:934593fa9dd04c2539b69e5c067547ac475b57b743c50836b866c620827a4a76
+oid sha256:e32e120c6d65576e6fe76fdf63d907877e27469586cc746d8cd6a71d07635ea9
 size 4472