Model save

README.md

@@ -0,0 +1,73 @@
+---
+license: mit
+base_model: HuggingFaceH4/mistral-7b-sft-beta
+tags:
+- generated_from_trainer
+model-index:
+- name: zephyr-7b-dpo-lora
+  results: []
+---
+
+<!-- This model card has been generated automatically according to the information the Trainer had access to. You
+should probably proofread and complete it, then remove this comment. -->
+
+# zephyr-7b-dpo-lora
+
+This model is a fine-tuned version of [HuggingFaceH4/mistral-7b-sft-beta](https://huggingface.co/HuggingFaceH4/mistral-7b-sft-beta) on the None dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.6941
+- Rewards/chosen: -0.0008
+- Rewards/rejected: 0.0011
+- Rewards/accuracies: 0.4870
+- Rewards/margins: -0.0019
+- Logps/rejected: -256.6144
+- Logps/chosen: -273.1258
+- Logits/rejected: -2.8923
+- Logits/chosen: -2.8913
+
+## Model description
+
+More information needed
+
+## Intended uses & limitations
+
+More information needed
+
+## Training and evaluation data
+
+More information needed
+
+## Training procedure
+
+### Training hyperparameters
+
+The following hyperparameters were used during training:
+- learning_rate: 5e-07
+- train_batch_size: 2
+- eval_batch_size: 4
+- seed: 42
+- distributed_type: multi-GPU
+- num_devices: 2
+- gradient_accumulation_steps: 32
+- total_train_batch_size: 128
+- total_eval_batch_size: 8
+- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
+- lr_scheduler_type: linear
+- lr_scheduler_warmup_ratio: 0.1
+- num_epochs: 3
+
+### Training results
+
+| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
+|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
+| 0.6931        | 0.84  | 4    | 0.6944          | -0.0020        | 0.0023           | 0.4680             | -0.0042         | -256.6025      | -273.1375    | -2.8922         | -2.8913       |
+| 0.6931        | 1.88  | 9    | 0.6937          | -0.0006        | 0.0011           | 0.4930             | -0.0016         | -256.6148      | -273.1236    | -2.8922         | -2.8914       |
+| 0.6934        | 2.51  | 12   | 0.6941          | -0.0008        | 0.0011           | 0.4870             | -0.0019         | -256.6144      | -273.1258    | -2.8923         | -2.8913       |
+
+
+### Framework versions
+
+- Transformers 4.35.0
+- Pytorch 2.1.0
+- Datasets 2.14.6
+- Tokenizers 0.14.1

adapter_config.json

@@ -0,0 +1,25 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "HuggingFaceH4/mistral-7b-sft-beta",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "lora_alpha": 16,
+  "lora_dropout": 0.1,
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 64,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "k_proj",
+    "q_proj",
+    "o_proj"
+  ],
+  "task_type": "CAUSAL_LM"
+}

adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b9aa4b56ad055076743bb6f6bf61d2cfad2be482bbca7691c5ffb99fcd147dca
+size 109086672

all_results.json

@@ -0,0 +1,21 @@
+{
+    "epoch": 2.51,
+    "eval_logits/chosen": -2.891324996948242,
+    "eval_logits/rejected": -2.892286777496338,
+    "eval_logps/chosen": -273.12579345703125,
+    "eval_logps/rejected": -256.6144104003906,
+    "eval_loss": 0.6940562725067139,
+    "eval_rewards/accuracies": 0.4869999885559082,
+    "eval_rewards/chosen": -0.0007941981311887503,
+    "eval_rewards/margins": -0.0018857381073758006,
+    "eval_rewards/rejected": 0.0010915396269410849,
+    "eval_runtime": 605.1943,
+    "eval_samples": 2000,
+    "eval_samples_per_second": 3.305,
+    "eval_steps_per_second": 0.413,
+    "train_loss": 0.6932857781648636,
+    "train_runtime": 2636.7323,
+    "train_samples": 611,
+    "train_samples_per_second": 0.695,
+    "train_steps_per_second": 0.005
+}

eval_results.json

@@ -0,0 +1,16 @@
+{
+    "epoch": 2.51,
+    "eval_logits/chosen": -2.891324996948242,
+    "eval_logits/rejected": -2.892286777496338,
+    "eval_logps/chosen": -273.12579345703125,
+    "eval_logps/rejected": -256.6144104003906,
+    "eval_loss": 0.6940562725067139,
+    "eval_rewards/accuracies": 0.4869999885559082,
+    "eval_rewards/chosen": -0.0007941981311887503,
+    "eval_rewards/margins": -0.0018857381073758006,
+    "eval_rewards/rejected": 0.0010915396269410849,
+    "eval_runtime": 605.1943,
+    "eval_samples": 2000,
+    "eval_samples_per_second": 3.305,
+    "eval_steps_per_second": 0.413
+}

runs/Jan08_17-05-18_jupyter-dli/events.out.tfevents.1704733679.jupyter-dli.4914.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:154097405d3bb0d0092bd997bc0d592672b62cd9819e9f4577538402db776720
+size 4985

runs/Jan08_17-18-29_jupyter-dli/events.out.tfevents.1704734332.jupyter-dli.9079.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3eb783e943d23929ef3b9dbef153c5967011abefe4a353d5aea24ae824f1c2fb
+size 8137

runs/Jan08_17-18-29_jupyter-dli/events.out.tfevents.1704737574.jupyter-dli.9079.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e5aa7982a4906e2c5e084afd4aa1c918b897d4da4d01047f4d4f963fce2809fb
+size 815

special_tokens_map.json

@@ -0,0 +1,35 @@
+{
+  "additional_special_tokens": [
+    "<unk>",
+    "<s>",
+    "</s>"
+  ],
+  "bos_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer_config.json

@@ -0,0 +1,46 @@
+{
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<unk>",
+    "<s>",
+    "</s>"
+  ],
+  "bos_token": "<s>",
+  "chat_template": "{% for message in messages %}\n{% if message['role'] == 'user' %}\n{{ '<|user|>\n' + message['content'] + eos_token }}\n{% elif message['role'] == 'system' %}\n{{ '<|system|>\n' + message['content'] + eos_token }}\n{% elif message['role'] == 'assistant' %}\n{{ '<|assistant|>\n'  + message['content'] + eos_token }}\n{% endif %}\n{% if loop.last and add_generation_prompt %}\n{{ '<|assistant|>' }}\n{% endif %}\n{% endfor %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "</s>",
+  "legacy": true,
+  "model_max_length": 2048,
+  "pad_token": "</s>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "LlamaTokenizer",
+  "truncation_side": "left",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": true
+}

train_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 2.51,
+    "train_loss": 0.6932857781648636,
+    "train_runtime": 2636.7323,
+    "train_samples": 611,
+    "train_samples_per_second": 0.695,
+    "train_steps_per_second": 0.005
+}

trainer_state.json

@@ -0,0 +1,104 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.5098039215686274,
+  "eval_steps": 100,
+  "global_step": 12,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.5e-07,
+      "logits/chosen": -2.9307734966278076,
+      "logits/rejected": -2.9072206020355225,
+      "logps/chosen": -257.9442138671875,
+      "logps/rejected": -223.16062927246094,
+      "loss": 0.6931,
+      "rewards/accuracies": 0.0,
+      "rewards/chosen": 0.0,
+      "rewards/margins": 0.0,
+      "rewards/rejected": 0.0,
+      "step": 1
+    },
+    {
+      "epoch": 0.84,
+      "eval_logits/chosen": -2.8913073539733887,
+      "eval_logits/rejected": -2.892223834991455,
+      "eval_logps/chosen": -273.1374816894531,
+      "eval_logps/rejected": -256.60247802734375,
+      "eval_loss": 0.6944335699081421,
+      "eval_rewards/accuracies": 0.46799999475479126,
+      "eval_rewards/chosen": -0.0019581823144108057,
+      "eval_rewards/margins": -0.004248426295816898,
+      "eval_rewards/rejected": 0.0022902432829141617,
+      "eval_runtime": 604.0376,
+      "eval_samples_per_second": 3.311,
+      "eval_steps_per_second": 0.414,
+      "step": 4
+    },
+    {
+      "epoch": 1.88,
+      "eval_logits/chosen": -2.891376495361328,
+      "eval_logits/rejected": -2.8922488689422607,
+      "eval_logps/chosen": -273.12359619140625,
+      "eval_logps/rejected": -256.61480712890625,
+      "eval_loss": 0.6936609148979187,
+      "eval_rewards/accuracies": 0.49300000071525574,
+      "eval_rewards/chosen": -0.0005687248194590211,
+      "eval_rewards/margins": -0.0016233286587521434,
+      "eval_rewards/rejected": 0.0010546041885390878,
+      "eval_runtime": 607.6521,
+      "eval_samples_per_second": 3.291,
+      "eval_steps_per_second": 0.411,
+      "step": 9
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1e-07,
+      "logits/chosen": -2.9432342052459717,
+      "logits/rejected": -2.949681520462036,
+      "logps/chosen": -269.6990051269531,
+      "logps/rejected": -258.2781982421875,
+      "loss": 0.6934,
+      "rewards/accuracies": 0.4340277910232544,
+      "rewards/chosen": 0.0009870771318674088,
+      "rewards/margins": 0.0012862730072811246,
+      "rewards/rejected": -0.0002991959627252072,
+      "step": 10
+    },
+    {
+      "epoch": 2.51,
+      "eval_logits/chosen": -2.891324996948242,
+      "eval_logits/rejected": -2.892286777496338,
+      "eval_logps/chosen": -273.12579345703125,
+      "eval_logps/rejected": -256.6144104003906,
+      "eval_loss": 0.6940562725067139,
+      "eval_rewards/accuracies": 0.4869999885559082,
+      "eval_rewards/chosen": -0.0007941981311887503,
+      "eval_rewards/margins": -0.0018857381073758006,
+      "eval_rewards/rejected": 0.0010915396269410849,
+      "eval_runtime": 604.8757,
+      "eval_samples_per_second": 3.306,
+      "eval_steps_per_second": 0.413,
+      "step": 12
+    },
+    {
+      "epoch": 2.51,
+      "step": 12,
+      "total_flos": 0.0,
+      "train_loss": 0.6932857781648636,
+      "train_runtime": 2636.7323,
+      "train_samples_per_second": 0.695,
+      "train_steps_per_second": 0.005
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 12,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "total_flos": 0.0,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c6e0a992426eb70b08db763df2e71fa42f4afe271e674e50c228443e1dc5bb93
+size 5752