Model save

README.md

@@ -0,0 +1,68 @@
+---
+base_model: meta-llama/Llama-3.2-3B-Instruct
+library_name: transformers
+model_name: Llama-3.2-3B-Instruct-cot-23-11
+tags:
+- generated_from_trainer
+- trl
+- dpo
+licence: license
+---
+
+# Model Card for Llama-3.2-3B-Instruct-cot-23-11
+
+This model is a fine-tuned version of [meta-llama/Llama-3.2-3B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct).
+It has been trained using [TRL](https://github.com/huggingface/trl).
+
+## Quick start
+
+```python
+from transformers import pipeline
+
+question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
+generator = pipeline("text-generation", model="Rudra-ai/Llama-3.2-3B-Instruct-cot-23-11", device="cuda")
+output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
+print(output["generated_text"])
+```
+
+## Training procedure
+
+
+
+This model was trained with DPO, a method introduced in [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://huggingface.co/papers/2305.18290).
+
+### Framework versions
+
+- TRL: 0.12.1
+- Transformers: 4.46.1
+- Pytorch: 2.4.1+cu124
+- Datasets: 3.1.0
+- Tokenizers: 0.20.3
+
+## Citations
+
+Cite DPO as:
+
+```bibtex
+@inproceedings{rafailov2023direct,
+    title        = {{Direct Preference Optimization: Your Language Model is Secretly a Reward Model}},
+    author       = {Rafael Rafailov and Archit Sharma and Eric Mitchell and Christopher D. Manning and Stefano Ermon and Chelsea Finn},
+    year         = 2023,
+    booktitle    = {Advances in Neural Information Processing Systems 36: Annual Conference on Neural Information Processing Systems 2023, NeurIPS 2023, New Orleans, LA, USA, December 10 - 16, 2023},
+    url          = {http://papers.nips.cc/paper_files/paper/2023/hash/a85b405ed65c6477a4fe8302b5e06ce7-Abstract-Conference.html},
+    editor       = {Alice Oh and Tristan Naumann and Amir Globerson and Kate Saenko and Moritz Hardt and Sergey Levine},
+}
+```
+
+Cite TRL as:
+    
+```bibtex
+@misc{vonwerra2022trl,
+	title        = {{TRL: Transformer Reinforcement Learning}},
+	author       = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallouédec},
+	year         = 2020,
+	journal      = {GitHub repository},
+	publisher    = {GitHub},
+	howpublished = {\url{https://github.com/huggingface/trl}}
+}
+```

all_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 1.0,
+    "total_flos": 0.0,
+    "train_loss": 0.08870122274137147,
+    "train_runtime": 1811.5821,
+    "train_samples": 2000,
+    "train_samples_per_second": 1.104,
+    "train_steps_per_second": 0.069
+}

generation_config.json

@@ -0,0 +1,12 @@
+{
+  "bos_token_id": 128000,
+  "do_sample": true,
+  "eos_token_id": [
+    128001,
+    128008,
+    128009
+  ],
+  "temperature": 0.6,
+  "top_p": 0.9,
+  "transformers_version": "4.46.1"
+}

train_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 1.0,
+    "total_flos": 0.0,
+    "train_loss": 0.08870122274137147,
+    "train_runtime": 1811.5821,
+    "train_samples": 2000,
+    "train_samples_per_second": 1.104,
+    "train_steps_per_second": 0.069
+}

trainer_state.json

@@ -0,0 +1,237 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 125,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.008,
+      "grad_norm": 477.2496337890625,
+      "learning_rate": 7.692307692307694e-07,
+      "logits/chosen": 2.15625,
+      "logits/rejected": 1.4375,
+      "logps/chosen": -146.0,
+      "logps/rejected": -131.0,
+      "loss": 2.7656,
+      "rewards/accuracies": 0.0,
+      "rewards/chosen": 0.0,
+      "rewards/margins": 0.0,
+      "rewards/rejected": 0.0,
+      "step": 1
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.534887996967882e-05,
+      "learning_rate": 7.692307692307694e-06,
+      "logits/chosen": 1.8359375,
+      "logits/rejected": 1.1328125,
+      "logps/chosen": -125.0,
+      "logps/rejected": -187.0,
+      "loss": 0.9163,
+      "rewards/accuracies": 0.8055555820465088,
+      "rewards/chosen": 1.46875,
+      "rewards/margins": 8.3125,
+      "rewards/rejected": -6.84375,
+      "step": 10
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 6.47921126473161e-12,
+      "learning_rate": 9.903926402016153e-06,
+      "logits/chosen": 0.3828125,
+      "logits/rejected": 0.357421875,
+      "logps/chosen": -160.0,
+      "logps/rejected": -540.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -2.921875,
+      "rewards/margins": 39.25,
+      "rewards/rejected": -42.25,
+      "step": 20
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.994430991895002e-15,
+      "learning_rate": 9.442228179894362e-06,
+      "logits/chosen": -0.060791015625,
+      "logits/rejected": -0.28515625,
+      "logps/chosen": -264.0,
+      "logps/rejected": -728.0,
+      "loss": 0.0075,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -13.9375,
+      "rewards/margins": 47.25,
+      "rewards/rejected": -61.25,
+      "step": 30
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.643975006320093e-14,
+      "learning_rate": 8.633301610170136e-06,
+      "logits/chosen": 1.0703125,
+      "logits/rejected": 0.59375,
+      "logps/chosen": -164.0,
+      "logps/rejected": -740.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -2.609375,
+      "rewards/margins": 59.25,
+      "rewards/rejected": -62.0,
+      "step": 40
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0163520159099946e-11,
+      "learning_rate": 7.540376726232648e-06,
+      "logits/chosen": 1.234375,
+      "logits/rejected": 0.6640625,
+      "logps/chosen": -130.0,
+      "logps/rejected": -636.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.015625,
+      "rewards/margins": 52.0,
+      "rewards/rejected": -53.0,
+      "step": 50
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.786671276360566e-15,
+      "learning_rate": 6.248882390836135e-06,
+      "logits/chosen": 1.359375,
+      "logits/rejected": 0.78515625,
+      "logps/chosen": -146.0,
+      "logps/rejected": -660.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.71875,
+      "rewards/margins": 52.5,
+      "rewards/rejected": -54.0,
+      "step": 60
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9845672956549099e-13,
+      "learning_rate": 4.859768718620656e-06,
+      "logits/chosen": 1.359375,
+      "logits/rejected": 0.83984375,
+      "logps/chosen": -150.0,
+      "logps/rejected": -640.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -0.58203125,
+      "rewards/margins": 52.0,
+      "rewards/rejected": -52.5,
+      "step": 70
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.1451468355701492e-15,
+      "learning_rate": 3.4816162744519266e-06,
+      "logits/chosen": 1.3125,
+      "logits/rejected": 0.73828125,
+      "logps/chosen": -149.0,
+      "logps/rejected": -688.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -0.97265625,
+      "rewards/margins": 55.75,
+      "rewards/rejected": -56.75,
+      "step": 80
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.8802392639811344e-13,
+      "learning_rate": 2.2221488349019903e-06,
+      "logits/chosen": 1.3359375,
+      "logits/rejected": 0.79296875,
+      "logps/chosen": -144.0,
+      "logps/rejected": -680.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.796875,
+      "rewards/margins": 54.5,
+      "rewards/rejected": -56.25,
+      "step": 90
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.8757966674675117e-07,
+      "learning_rate": 1.1798131208919628e-06,
+      "logits/chosen": 1.359375,
+      "logits/rejected": 0.76171875,
+      "logps/chosen": -150.0,
+      "logps/rejected": -632.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.4453125,
+      "rewards/margins": 49.75,
+      "rewards/rejected": -51.25,
+      "step": 100
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.703149137021777e-14,
+      "learning_rate": 4.3608367469340553e-07,
+      "logits/chosen": 1.34375,
+      "logits/rejected": 0.7734375,
+      "logps/chosen": -153.0,
+      "logps/rejected": -628.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -2.078125,
+      "rewards/margins": 50.0,
+      "rewards/rejected": -52.0,
+      "step": 110
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 7.835807033238558e-16,
+      "learning_rate": 4.909437331777178e-08,
+      "logits/chosen": 1.3515625,
+      "logits/rejected": 0.7734375,
+      "logps/chosen": -144.0,
+      "logps/rejected": -676.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.203125,
+      "rewards/margins": 54.0,
+      "rewards/rejected": -55.25,
+      "step": 120
+    },
+    {
+      "epoch": 1.0,
+      "step": 125,
+      "total_flos": 0.0,
+      "train_loss": 0.08870122274137147,
+      "train_runtime": 1811.5821,
+      "train_samples_per_second": 1.104,
+      "train_steps_per_second": 0.069
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 125,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 200,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}