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
+## 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

checkpoint-200/adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d17c363cd05be312b15770b466c1338fec2b141ac20ae3fe7485fd24b8318d8c
+oid sha256:dff98b7eb0d073a715b62e6df5759c50443fc67bdcecc94f7ecfdc034dcdba12
 size 19690328

checkpoint-200/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0962560544030656e61dc01e1d9860980dbf6790ba53119ffc396e82401a5857
+oid sha256:39ceec45f3a10d8fa37d8a0c5bc8ee99ffac785fa4c0f43e5f969dedd176fad4
 size 38087098

checkpoint-200/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6b698a3ffed4744d0125b17418b2a3d7497f5d9d4c57f4d5e997a2cb6ffd50f2
+oid sha256:eb1952519fd1553759d558392a47f35c529236b7568c03d6c6851620ee48a413
 size 14244

checkpoint-200/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f5b43ac6655c6879b737216c959393eb2c7f9aec4594023f8e587c864aeb04e5
+oid sha256:1bc034746d61c122e35ebcec68ba024674bd5039ad1df2f10fb31340e9444bca
 size 1064

checkpoint-200/trainer_state.json

@@ -1,7 +1,7 @@
 {
-  "best_metric": 0.4239814877510071,
-  "best_model_checkpoint": "output/checkpoint-50",
-  "epoch": 2.888086642599278,
+  "best_metric": 0.4308854341506958,
+  "best_model_checkpoint": "output/checkpoint-100",
+  "epoch": 0.5414551607445008,
   "eval_steps": 50,
   "global_step": 200,
   "is_hyper_param_search": false,
@@ -9,77 +9,77 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.01,
+      "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 0.3991,
+      "loss": 0.9268,
       "step": 1
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.8902439024390243e-05,
-      "loss": 0.3172,
+      "epoch": 0.14,
+      "learning_rate": 6.111111111111112e-06,
+      "loss": 0.5897,
       "step": 50
     },
     {
-      "epoch": 0.72,
-      "eval_accuracy": 0.84,
-      "eval_loss": 0.4239814877510071,
-      "eval_runtime": 28.6347,
-      "eval_samples_per_second": 6.111,
-      "eval_steps_per_second": 1.537,
+      "epoch": 0.14,
+      "eval_accuracy": 0.8628571428571429,
+      "eval_loss": 0.4395545721054077,
+      "eval_runtime": 26.2003,
+      "eval_samples_per_second": 6.679,
+      "eval_steps_per_second": 1.679,
       "step": 50
     },
     {
-      "epoch": 1.44,
-      "learning_rate": 1.2804878048780488e-05,
-      "loss": 0.1954,
+      "epoch": 0.27,
+      "learning_rate": 1.3055555555555557e-05,
+      "loss": 0.4516,
       "step": 100
     },
     {
-      "epoch": 1.44,
-      "eval_accuracy": 0.8514285714285714,
-      "eval_loss": 0.5004297494888306,
-      "eval_runtime": 28.6111,
-      "eval_samples_per_second": 6.117,
-      "eval_steps_per_second": 1.538,
+      "epoch": 0.27,
+      "eval_accuracy": 0.8457142857142858,
+      "eval_loss": 0.4308854341506958,
+      "eval_runtime": 28.5089,
+      "eval_samples_per_second": 6.138,
+      "eval_steps_per_second": 1.543,
       "step": 100
     },
     {
-      "epoch": 2.17,
-      "learning_rate": 6.707317073170733e-06,
-      "loss": 0.1457,
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.4086,
       "step": 150
     },
     {
-      "epoch": 2.17,
-      "eval_accuracy": 0.8628571428571429,
-      "eval_loss": 0.5306527614593506,
-      "eval_runtime": 28.6321,
-      "eval_samples_per_second": 6.112,
-      "eval_steps_per_second": 1.537,
+      "epoch": 0.41,
+      "eval_accuracy": 0.8114285714285714,
+      "eval_loss": 0.45158952474594116,
+      "eval_runtime": 28.4711,
+      "eval_samples_per_second": 6.147,
+      "eval_steps_per_second": 1.545,
       "step": 150
     },
     {
-      "epoch": 2.89,
-      "learning_rate": 6.097560975609757e-07,
-      "loss": 0.1312,
+      "epoch": 0.54,
+      "learning_rate": 1.8475609756097565e-05,
+      "loss": 0.4227,
       "step": 200
     },
     {
-      "epoch": 2.89,
-      "eval_accuracy": 0.8514285714285714,
-      "eval_loss": 0.6020106077194214,
-      "eval_runtime": 28.6498,
-      "eval_samples_per_second": 6.108,
-      "eval_steps_per_second": 1.536,
+      "epoch": 0.54,
+      "eval_accuracy": 0.8114285714285714,
+      "eval_loss": 0.4496140480041504,
+      "eval_runtime": 26.1937,
+      "eval_samples_per_second": 6.681,
+      "eval_steps_per_second": 1.68,
       "step": 200
     }
   ],
   "logging_steps": 50,
-  "max_steps": 200,
+  "max_steps": 800,
   "num_train_epochs": 3,
   "save_steps": 50,
-  "total_flos": 7.64381811965952e+16,
+  "total_flos": 7.441466063671296e+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:d6e1226be91d4f2bcdcde81b49645b5882b397e67c3f85a88d2c0a49245bf8c5
+oid sha256:eab90039203ebde7fc5537527df9e43bcc8971062162dee256a3fd06a20fcd44
 size 4600