Training in progress, step 350, checkpoint

checkpoint-350/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-350/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7555e30aa42692b57cadf4f85c3747eca4196876d5bec8211bf4ebe2255728ae
+oid sha256:072a1bb016febd596c985d21de00e4d37069e78d3e8fc3a1377a65c7f74d2b88
 size 9873829

checkpoint-350/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c13a98ef9b2d51b7ea0b0b2e95cd4e22374b9af80a47bb46d7251336048aa398
+oid sha256:179bdfcba6dd304058e7e6ca1a883eb8da929965e94dcefc374651047da060ef
 size 42788

checkpoint-350/trainer_state.json

@@ -1,6 +1,6 @@
 {
-  "best_metric": 0.622471034526825,
-  "best_model_checkpoint": "output/checkpoint-300",
+  "best_metric": 0.6430063247680664,
+  "best_model_checkpoint": "output/checkpoint-50",
   "epoch": 0.2805892373985369,
   "eval_steps": 50,
   "global_step": 350,
@@ -11,112 +11,112 @@
     {
       "epoch": 0.0,
       "learning_rate": 4.444444444444445e-07,
-      "loss": 0.5558,
+      "loss": 0.5542,
       "step": 1
     },
     {
       "epoch": 0.04,
       "learning_rate": 2e-05,
-      "loss": 0.5679,
+      "loss": 0.554,
       "step": 50
     },
     {
       "epoch": 0.04,
-      "eval_accuracy": 0.6776119402985075,
-      "eval_loss": 0.6410807371139526,
-      "eval_runtime": 61.731,
-      "eval_samples_per_second": 5.427,
-      "eval_steps_per_second": 1.361,
+      "eval_accuracy": 0.6626865671641791,
+      "eval_loss": 0.6430063247680664,
+      "eval_runtime": 61.7692,
+      "eval_samples_per_second": 5.423,
+      "eval_steps_per_second": 1.36,
       "step": 50
     },
     {
       "epoch": 0.08,
       "learning_rate": 4.222222222222222e-05,
-      "loss": 0.6108,
+      "loss": 0.5957,
       "step": 100
     },
     {
       "epoch": 0.08,
-      "eval_accuracy": 0.6925373134328359,
-      "eval_loss": 0.6733226776123047,
-      "eval_runtime": 56.5238,
-      "eval_samples_per_second": 5.927,
-      "eval_steps_per_second": 1.486,
+      "eval_accuracy": 0.6686567164179105,
+      "eval_loss": 0.6705919504165649,
+      "eval_runtime": 60.4388,
+      "eval_samples_per_second": 5.543,
+      "eval_steps_per_second": 1.39,
       "step": 100
     },
     {
       "epoch": 0.12,
       "learning_rate": 6.444444444444446e-05,
-      "loss": 0.5865,
+      "loss": 0.5668,
       "step": 150
     },
     {
       "epoch": 0.12,
-      "eval_accuracy": 0.6865671641791045,
-      "eval_loss": 0.7122625708580017,
-      "eval_runtime": 61.5948,
-      "eval_samples_per_second": 5.439,
-      "eval_steps_per_second": 1.364,
+      "eval_accuracy": 0.6805970149253732,
+      "eval_loss": 0.7224913835525513,
+      "eval_runtime": 56.4882,
+      "eval_samples_per_second": 5.93,
+      "eval_steps_per_second": 1.487,
       "step": 150
     },
     {
       "epoch": 0.16,
       "learning_rate": 8.666666666666667e-05,
-      "loss": 0.5752,
+      "loss": 0.5626,
       "step": 200
     },
     {
       "epoch": 0.16,
-      "eval_accuracy": 0.6835820895522388,
-      "eval_loss": 0.6697484254837036,
-      "eval_runtime": 61.5178,
-      "eval_samples_per_second": 5.446,
-      "eval_steps_per_second": 1.365,
+      "eval_accuracy": 0.6746268656716418,
+      "eval_loss": 0.7115103006362915,
+      "eval_runtime": 56.506,
+      "eval_samples_per_second": 5.929,
+      "eval_steps_per_second": 1.487,
       "step": 200
     },
     {
       "epoch": 0.2,
       "learning_rate": 9.80430528375734e-05,
-      "loss": 0.5826,
+      "loss": 0.5604,
       "step": 250
     },
     {
       "epoch": 0.2,
-      "eval_accuracy": 0.6805970149253732,
-      "eval_loss": 0.6812018752098083,
-      "eval_runtime": 61.5335,
-      "eval_samples_per_second": 5.444,
-      "eval_steps_per_second": 1.365,
+      "eval_accuracy": 0.6746268656716418,
+      "eval_loss": 0.700947105884552,
+      "eval_runtime": 56.4927,
+      "eval_samples_per_second": 5.93,
+      "eval_steps_per_second": 1.487,
       "step": 250
     },
     {
       "epoch": 0.24,
       "learning_rate": 9.315068493150684e-05,
-      "loss": 0.593,
+      "loss": 0.5675,
       "step": 300
     },
     {
       "epoch": 0.24,
-      "eval_accuracy": 0.6865671641791045,
-      "eval_loss": 0.622471034526825,
-      "eval_runtime": 61.6449,
-      "eval_samples_per_second": 5.434,
-      "eval_steps_per_second": 1.363,
+      "eval_accuracy": 0.6597014925373135,
+      "eval_loss": 0.662155032157898,
+      "eval_runtime": 56.4903,
+      "eval_samples_per_second": 5.93,
+      "eval_steps_per_second": 1.487,
       "step": 300
     },
     {
       "epoch": 0.28,
       "learning_rate": 8.825831702544032e-05,
-      "loss": 0.5945,
+      "loss": 0.5725,
       "step": 350
     },
     {
       "epoch": 0.28,
-      "eval_accuracy": 0.6686567164179105,
-      "eval_loss": 0.6458173394203186,
-      "eval_runtime": 61.6102,
-      "eval_samples_per_second": 5.437,
-      "eval_steps_per_second": 1.363,
+      "eval_accuracy": 0.6567164179104478,
+      "eval_loss": 0.6746364831924438,
+      "eval_runtime": 56.482,
+      "eval_samples_per_second": 5.931,
+      "eval_steps_per_second": 1.487,
       "step": 350
     }
   ],

checkpoint-350/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e2cf806244b71c753bc4b76d2695973243d02590ea5a24f377343d4ee219e984
+oid sha256:87bd30bd4b21b4a0c66f1fd545858ef6e2d33b5984a8d292e79011acc3affd6b
 size 4472