Training in progress, step 500, checkpoint

checkpoint-500/README.md

@@ -0,0 +1,21 @@
+---
+library_name: peft
+---
+## Training procedure
+
+
+The following `bitsandbytes` quantization config was used during training:
+- quant_method: bitsandbytes
+- load_in_8bit: False
+- load_in_4bit: True
+- 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: nf4
+- bnb_4bit_use_double_quant: True
+- bnb_4bit_compute_dtype: bfloat16
+### Framework versions
+
+
+- PEFT 0.6.0.dev0

checkpoint-500/adapter_config.json

@@ -0,0 +1,21 @@
+{
+  "auto_mapping": null,
+  "base_model_name_or_path": "codellama/CodeLlama-7b-Instruct-hf",
+  "bias": "lora_only",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "lora_alpha": 8,
+  "lora_dropout": 0.1,
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "v_proj"
+  ],
+  "task_type": "CAUSAL_LM"
+}

checkpoint-500/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ad7708a2747ce9361a6a19d614cf2f8a789016ac9c1040a6983a128caf3597bf
+size 16822989

checkpoint-500/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9f0e75e4a45a393a1e590a6cf95a5d1a6b1b89be8617ef06333c6361700ab760
+size 8555781

checkpoint-500/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05300415ab1c9d5ed3069424316a4446d3455810c1b6392e6597ad110acc77e2
+size 14575

checkpoint-500/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:52a5b043e0c4fbe9dc95c92dc03d52e59908d2fabc81f7b2c54ca17d70b4332c
+size 627

checkpoint-500/trainer_state.json

@@ -0,0 +1,125 @@
+{
+  "best_metric": 0.23951096832752228,
+  "best_model_checkpoint": "./text2sql/codellama_instruct_spider_e10/checkpoint-500",
+  "epoch": 1.8475750577367207,
+  "eval_steps": 100,
+  "global_step": 500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.08641975308642e-07,
+      "loss": 1.3897,
+      "step": 1
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.54320987654321e-05,
+      "loss": 1.3449,
+      "step": 50
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.08641975308642e-05,
+      "loss": 0.822,
+      "step": 100
+    },
+    {
+      "epoch": 0.37,
+      "eval_loss": 0.5313276648521423,
+      "eval_runtime": 34.4984,
+      "eval_samples_per_second": 29.972,
+      "eval_steps_per_second": 3.768,
+      "step": 100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.62962962962963e-05,
+      "loss": 0.4342,
+      "step": 150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9251379038613084e-05,
+      "loss": 0.3014,
+      "step": 200
+    },
+    {
+      "epoch": 0.74,
+      "eval_loss": 0.2763093411922455,
+      "eval_runtime": 34.3952,
+      "eval_samples_per_second": 30.062,
+      "eval_steps_per_second": 3.78,
+      "step": 200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.826635145784082e-05,
+      "loss": 0.2401,
+      "step": 250
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.728132387706856e-05,
+      "loss": 0.2091,
+      "step": 300
+    },
+    {
+      "epoch": 1.11,
+      "eval_loss": 0.24685852229595184,
+      "eval_runtime": 34.5214,
+      "eval_samples_per_second": 29.952,
+      "eval_steps_per_second": 3.766,
+      "step": 300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.62962962962963e-05,
+      "loss": 0.1837,
+      "step": 350
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.5311268715524035e-05,
+      "loss": 0.1697,
+      "step": 400
+    },
+    {
+      "epoch": 1.48,
+      "eval_loss": 0.2400880753993988,
+      "eval_runtime": 34.3334,
+      "eval_samples_per_second": 30.116,
+      "eval_steps_per_second": 3.786,
+      "step": 400
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.432624113475177e-05,
+      "loss": 0.1589,
+      "step": 450
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.334121355397952e-05,
+      "loss": 0.1495,
+      "step": 500
+    },
+    {
+      "epoch": 1.85,
+      "eval_loss": 0.23951096832752228,
+      "eval_runtime": 34.3515,
+      "eval_samples_per_second": 30.101,
+      "eval_steps_per_second": 3.784,
+      "step": 500
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 2700,
+  "num_train_epochs": 10,
+  "save_steps": 100,
+  "total_flos": 1.529928488064123e+17,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-500/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f894492930ff690bb0d1428d042fd66b325480fe96ad5ff777a305a69efec314
+size 4219