Training in progress, step 150, checkpoint

checkpoint-150/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-150/adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:74219a40b2a483f29b2465150aee10b8960fec218f3d348ddeac47e82e2c4b03
+oid sha256:4c94f15ff2c42b45f0b98828b350fac0c17f29fb4e9dda241d7200e9e0cddc01
 size 19690328

checkpoint-150/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:43bb8da1a1d887a7c5534d9c0b1a6d2db9b04d4703693bd053f07a6f8d7d66b0
+oid sha256:b1cab13a86e8437b1373cd19d8ccc2d83d318ef0664e9eb6207879bb5cda4d65
 size 38087098

checkpoint-150/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e9d1ae2cf1bf15efc09451f1cba487634dc50fa186aa8156fca93af0393295a8
+oid sha256:7316f315e15d9f459afc4545b937fa78d98d84c01e3954b02e2f03060f7bb97c
 size 14244

checkpoint-150/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4ea969f99dfb78c951cacb5a375348af305c0ed94ae738a641b514e80ad1fd2b
+oid sha256:3babfedf2934ed05b943394d6fd07186959f49c5b3ea9a3f628bda5f318fdf7a
 size 1064

checkpoint-150/trainer_state.json

@@ -1,7 +1,7 @@
 {
-  "best_metric": 0.4239814877510071,
-  "best_model_checkpoint": "output/checkpoint-50",
-  "epoch": 2.1660649819494586,
+  "best_metric": 0.4308854341506958,
+  "best_model_checkpoint": "output/checkpoint-100",
+  "epoch": 0.40609137055837563,
   "eval_steps": 50,
   "global_step": 150,
   "is_hyper_param_search": false,
@@ -9,62 +9,62 @@
   "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
     }
   ],
   "logging_steps": 50,
-  "max_steps": 200,
+  "max_steps": 800,
   "num_train_epochs": 3,
   "save_steps": 50,
-  "total_flos": 5.719603163824128e+16,
+  "total_flos": 5.573993923547136e+16,
   "trial_name": null,
   "trial_params": null
 }

checkpoint-150/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d6e1226be91d4f2bcdcde81b49645b5882b397e67c3f85a88d2c0a49245bf8c5
+oid sha256:eab90039203ebde7fc5537527df9e43bcc8971062162dee256a3fd06a20fcd44
 size 4600