Model save

README.md

@@ -0,0 +1,69 @@
+---
+base_model: jkazdan/llama3-8b-stage-2
+library_name: transformers
+model_name: llama-8b-dpo-full-1
+tags:
+- generated_from_trainer
+- trl
+- dpo
+licence: license
+---
+
+# Model Card for llama-8b-dpo-full-1
+
+This model is a fine-tuned version of [jkazdan/llama3-8b-stage-2](https://huggingface.co/jkazdan/llama3-8b-stage-2).
+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="jkazdan/llama-8b-dpo-full-1", 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: 0.15.2
+- Transformers: 4.49.0
+- Pytorch: 2.6.0
+- Datasets: 3.3.2
+- Tokenizers: 0.21.0
+
+## 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": 0.685900236430921,
+    "train_runtime": 550.0055,
+    "train_samples": 4840,
+    "train_samples_per_second": 8.8,
+    "train_steps_per_second": 0.069
+}

generation_config.json

@@ -0,0 +1,9 @@
+{
+  "bos_token_id": 128000,
+  "do_sample": true,
+  "eos_token_id": 128001,
+  "max_length": 4096,
+  "temperature": 0.6,
+  "top_p": 0.9,
+  "transformers_version": "4.49.0"
+}

train_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 1.0,
+    "total_flos": 0.0,
+    "train_loss": 0.685900236430921,
+    "train_runtime": 550.0055,
+    "train_samples": 4840,
+    "train_samples_per_second": 8.8,
+    "train_steps_per_second": 0.069
+}

trainer_state.json

@@ -0,0 +1,102 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 100,
+  "global_step": 38,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02631578947368421,
+      "grad_norm": 2.592954402820238,
+      "learning_rate": 1.25e-07,
+      "logits/chosen": 0.4306640625,
+      "logits/rejected": 0.537109375,
+      "logps/chosen": -237.5,
+      "logps/rejected": -228.0,
+      "loss": 0.6914,
+      "rewards/accuracies": 0.0,
+      "rewards/chosen": 0.0,
+      "rewards/margins": 0.0,
+      "rewards/rejected": 0.0,
+      "step": 1
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 2.5257972591559534,
+      "learning_rate": 4.6255428393240354e-07,
+      "logits/chosen": 0.4337022602558136,
+      "logits/rejected": 0.4523654580116272,
+      "logps/chosen": -243.11111450195312,
+      "logps/rejected": -236.94444274902344,
+      "loss": 0.6919,
+      "rewards/accuracies": 0.28125,
+      "rewards/chosen": 0.004753430839627981,
+      "rewards/margins": 0.0010722478618845344,
+      "rewards/rejected": 0.0036805470008403063,
+      "step": 10
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 2.555743613102717,
+      "learning_rate": 2.730670898658255e-07,
+      "logits/chosen": 0.36152344942092896,
+      "logits/rejected": 0.45283204317092896,
+      "logps/chosen": -243.5,
+      "logps/rejected": -237.3000030517578,
+      "loss": 0.6873,
+      "rewards/accuracies": 0.54296875,
+      "rewards/chosen": 0.03550415113568306,
+      "rewards/margins": 0.011375045403838158,
+      "rewards/rejected": 0.02410888671875,
+      "step": 20
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 2.6433738154085615,
+      "learning_rate": 6.524777069483525e-08,
+      "logits/chosen": 0.35273438692092896,
+      "logits/rejected": 0.47441405057907104,
+      "logps/chosen": -243.14999389648438,
+      "logps/rejected": -235.14999389648438,
+      "loss": 0.6832,
+      "rewards/accuracies": 0.594531238079071,
+      "rewards/chosen": 0.06180419772863388,
+      "rewards/margins": 0.02066650427877903,
+      "rewards/rejected": 0.04110107570886612,
+      "step": 30
+    },
+    {
+      "epoch": 1.0,
+      "step": 38,
+      "total_flos": 0.0,
+      "train_loss": 0.685900236430921,
+      "train_runtime": 550.0055,
+      "train_samples_per_second": 8.8,
+      "train_steps_per_second": 0.069
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 38,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 100,
+  "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": 8,
+  "trial_name": null,
+  "trial_params": null
+}