Upload SFT artifact (adapter_only) from phase2-codellama-7b-lora-kaggle-1h

adapter_config.json

@@ -0,0 +1,43 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "vodkaslime/codellama-7b-hf",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "o_proj",
+    "q_proj",
+    "v_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:77efbe297b950aec2f560dc4136e44cb82803c3bbec332aa4cfd4f51374e0934
+size 67143296

phase2_report.json

@@ -0,0 +1,15 @@
+{
+  "run_dir": "/kaggle/working/python-llm-grpo-new-techniques/training/output/phase2-codellama-7b-lora-kaggle-1h",
+  "trainer_state_path": "/kaggle/working/python-llm-grpo-new-techniques/training/output/phase2-codellama-7b-lora-kaggle-1h/checkpoint-43/trainer_state.json",
+  "metrics": {
+    "global_step": 43,
+    "best_metric": null,
+    "last_train_loss": 0.7746126174926757,
+    "best_eval_loss": null,
+    "last_eval_loss": null,
+    "last_learning_rate": 0.00015600000000000002,
+    "last_step_logged": 40,
+    "num_train_loss_logs": 4,
+    "num_eval_loss_logs": 0
+  }
+}

phase2_report.md

@@ -0,0 +1,15 @@
+# Phase 2 SFT Report
+
+- Run directory: `/kaggle/working/python-llm-grpo-new-techniques/training/output/phase2-codellama-7b-lora-kaggle-1h`
+- Trainer state: `/kaggle/working/python-llm-grpo-new-techniques/training/output/phase2-codellama-7b-lora-kaggle-1h/checkpoint-43/trainer_state.json`
+
+## Metrics
+- global_step: `43`
+- best_metric: `None`
+- last_train_loss: `0.7746126174926757`
+- best_eval_loss: `None`
+- last_eval_loss: `None`
+- last_learning_rate: `0.00015600000000000002`
+- last_step_logged: `40`
+- num_train_loss_logs: `4`
+- num_eval_loss_logs: `0`

tokenizer_config.json

@@ -0,0 +1,13 @@
+{
+  "backend": "tokenizers",
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "</s>",
+  "is_local": false,
+  "legacy": null,
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "</s>",
+  "sp_model_kwargs": {},
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": "<unk>"
+}

training_manifest.json

@@ -0,0 +1,41 @@
+{
+  "started_at_utc": "2026-03-19T11:54:34.606942+00:00",
+  "ended_at_utc": "2026-03-19T13:30:48.781371+00:00",
+  "train_elapsed_sec": 5774.174,
+  "stopped_by_time_limit": false,
+  "global_step": 43,
+  "train_metrics": {
+    "train_runtime": 5773.5801,
+    "train_samples_per_second": 0.118,
+    "train_steps_per_second": 0.007,
+    "total_flos": 2.771601971163955e+16,
+    "train_loss": 0.9663696178170138,
+    "epoch": 1.0
+  },
+  "args": {
+    "model_name": "vodkaslime/codellama-7b-hf",
+    "train_file": "/kaggle/working/python-llm-grpo-new-techniques/training/data/train.jsonl",
+    "val_file": "/kaggle/working/python-llm-grpo-new-techniques/training/data/val.jsonl",
+    "output_dir": "/kaggle/working/python-llm-grpo-new-techniques/training/output/phase2-codellama-7b-lora-kaggle-1h",
+    "block_size": 1024,
+    "num_train_epochs": 1,
+    "max_steps": -1,
+    "max_runtime_minutes": 55.0,
+    "max_runtime_seconds": 7200.0,
+    "effective_max_runtime_seconds": 7200.0,
+    "per_device_train_batch_size": 1,
+    "per_device_eval_batch_size": 1,
+    "gradient_accumulation_steps": 16,
+    "learning_rate": 0.0002,
+    "save_steps": 100,
+    "eval_steps": 100,
+    "save_total_limit": 2,
+    "warmup_steps": 50,
+    "use_4bit": true,
+    "bf16": true,
+    "fp16": false,
+    "gradient_checkpointing": true,
+    "seed": 42,
+    "artifact_mode": "adapter_only"
+  }
+}

upload_manifest.json

@@ -0,0 +1,14 @@
+{
+  "generated_at_utc": "2026-03-19T13:40:53.857341+00:00",
+  "upload_mode": "adapter_only",
+  "source_run_dir": "/kaggle/working/python-llm-grpo-new-techniques/training/output/phase2-codellama-7b-lora-kaggle-1h",
+  "included_files": [
+    "adapter_config.json",
+    "adapter_model.safetensors",
+    "phase2_report.json",
+    "phase2_report.md",
+    "tokenizer.json",
+    "tokenizer_config.json",
+    "training_manifest.json"
+  ]
+}