fine tune swag

README.md

@@ -0,0 +1,74 @@
+---
+language:
+- en
+license: mit
+base_model: microsoft/mdeberta-v3-base
+tags:
+- generated_from_trainer
+datasets:
+- swag
+metrics:
+- accuracy
+model-index:
+- name: llm-mdeberta-v3-swag
+  results:
+  - task:
+      name: Multiple Choice
+      type: multiple-choice
+    dataset:
+      name: SWAG
+      type: swag
+      config: regular
+      split: validation
+      args: regular
+    metrics:
+    - name: Accuracy
+      type: accuracy
+      value: 0.777816653251648
+---
+
+<!-- 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. -->
+
+# llm-mdeberta-v3-swag
+
+This model is a fine-tuned version of [microsoft/mdeberta-v3-base](https://huggingface.co/microsoft/mdeberta-v3-base) on the SWAG dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.8202
+- Accuracy: 0.7778
+
+## 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-05
+- train_batch_size: 8
+- eval_batch_size: 8
+- seed: 42
+- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
+- lr_scheduler_type: linear
+- num_epochs: 3.0
+
+### Training results
+
+
+
+### Framework versions
+
+- Transformers 4.37.0.dev0
+- Pytorch 2.1.0+cu121
+- Datasets 2.15.0
+- Tokenizers 0.15.0

added_tokens.json

@@ -0,0 +1,3 @@
+{
+  "[MASK]": 250101
+}

all_results.json

@@ -0,0 +1,14 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.777816653251648,
+    "eval_loss": 0.8202171921730042,
+    "eval_runtime": 366.8314,
+    "eval_samples": 20006,
+    "eval_samples_per_second": 54.537,
+    "eval_steps_per_second": 6.818,
+    "train_loss": 0.6582008492790169,
+    "train_runtime": 15084.1881,
+    "train_samples": 73546,
+    "train_samples_per_second": 14.627,
+    "train_steps_per_second": 1.829
+}

config.json

@@ -0,0 +1,35 @@
+{
+  "_name_or_path": "microsoft/mdeberta-v3-base",
+  "architectures": [
+    "DebertaV2ForMultipleChoice"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-07,
+  "max_position_embeddings": 512,
+  "max_relative_positions": -1,
+  "model_type": "deberta-v2",
+  "norm_rel_ebd": "layer_norm",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "pad_token_id": 0,
+  "pooler_dropout": 0,
+  "pooler_hidden_act": "gelu",
+  "pooler_hidden_size": 768,
+  "pos_att_type": [
+    "p2c",
+    "c2p"
+  ],
+  "position_biased_input": false,
+  "position_buckets": 256,
+  "relative_attention": true,
+  "share_att_key": true,
+  "torch_dtype": "float32",
+  "transformers_version": "4.37.0.dev0",
+  "type_vocab_size": 0,
+  "vocab_size": 251000
+}

eval_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.777816653251648,
+    "eval_loss": 0.8202171921730042,
+    "eval_runtime": 366.8314,
+    "eval_samples": 20006,
+    "eval_samples_per_second": 54.537,
+    "eval_steps_per_second": 6.818
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:49a3c54108f314f93a59b05591eb83b27b3624be5ed7914ffb4b87c9e5d732db
+size 1115265124

runs/Dec17_22-15-06_a41ddc482777/events.out.tfevents.1702851476.a41ddc482777.2398.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:18b5f5614e5eb5a568387966334bb4c4ddf967f790d777e7dcff32fce94ab062
+size 13582

runs/Dec17_22-15-06_a41ddc482777/events.out.tfevents.1702866932.a41ddc482777.2398.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:86176a3404a512a677dc548fbe6b079ba99046255a9c8919f081dac975b54bb4
+size 417

special_tokens_map.json

@@ -0,0 +1,15 @@
+{
+  "bos_token": "[CLS]",
+  "cls_token": "[CLS]",
+  "eos_token": "[SEP]",
+  "mask_token": "[MASK]",
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "unk_token": {
+    "content": "[UNK]",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  }
+}

spm.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:13c8d666d62a7bc4ac8f040aab68e942c861f93303156cc28f5c7e885d86d6e3
+size 4305025

tokenizer_config.json

@@ -0,0 +1,58 @@
+{
+  "added_tokens_decoder": {
+    "0": {
+      "content": "[PAD]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "[CLS]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "[SEP]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "[UNK]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "250101": {
+      "content": "[MASK]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "bos_token": "[CLS]",
+  "clean_up_tokenization_spaces": true,
+  "cls_token": "[CLS]",
+  "do_lower_case": false,
+  "eos_token": "[SEP]",
+  "mask_token": "[MASK]",
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "sp_model_kwargs": {},
+  "split_by_punct": false,
+  "tokenizer_class": "DebertaV2Tokenizer",
+  "unk_token": "[UNK]",
+  "vocab_type": "spm"
+}

train_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 3.0,
+    "train_loss": 0.6582008492790169,
+    "train_runtime": 15084.1881,
+    "train_samples": 73546,
+    "train_samples_per_second": 14.627,
+    "train_steps_per_second": 1.829
+}

trainer_state.json

@@ -0,0 +1,360 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 27582,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.909361177579581e-05,
+      "loss": 1.1223,
+      "step": 500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.818722355159162e-05,
+      "loss": 1.035,
+      "step": 1000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.728083532738743e-05,
+      "loss": 1.0383,
+      "step": 1500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.637444710318324e-05,
+      "loss": 0.9976,
+      "step": 2000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.546805887897905e-05,
+      "loss": 0.9862,
+      "step": 2500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4561670654774854e-05,
+      "loss": 0.9828,
+      "step": 3000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3655282430570665e-05,
+      "loss": 0.991,
+      "step": 3500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2748894206366476e-05,
+      "loss": 0.9492,
+      "step": 4000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.184250598216228e-05,
+      "loss": 0.9492,
+      "step": 4500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.093611775795809e-05,
+      "loss": 0.9483,
+      "step": 5000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.00297295337539e-05,
+      "loss": 0.9157,
+      "step": 5500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.912334130954971e-05,
+      "loss": 0.9092,
+      "step": 6000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.821695308534552e-05,
+      "loss": 0.8941,
+      "step": 6500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.731056486114133e-05,
+      "loss": 0.9126,
+      "step": 7000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.640417663693714e-05,
+      "loss": 0.8869,
+      "step": 7500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.549778841273294e-05,
+      "loss": 0.8928,
+      "step": 8000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.459140018852875e-05,
+      "loss": 0.8764,
+      "step": 8500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.368501196432456e-05,
+      "loss": 0.8788,
+      "step": 9000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.277862374012037e-05,
+      "loss": 0.7333,
+      "step": 9500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.187223551591618e-05,
+      "loss": 0.6604,
+      "step": 10000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.0965847291711983e-05,
+      "loss": 0.6689,
+      "step": 10500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0059459067507794e-05,
+      "loss": 0.6555,
+      "step": 11000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9153070843303605e-05,
+      "loss": 0.658,
+      "step": 11500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8246682619099413e-05,
+      "loss": 0.6627,
+      "step": 12000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.734029439489522e-05,
+      "loss": 0.6626,
+      "step": 12500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.643390617069103e-05,
+      "loss": 0.6381,
+      "step": 13000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.552751794648684e-05,
+      "loss": 0.6319,
+      "step": 13500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.462112972228265e-05,
+      "loss": 0.6766,
+      "step": 14000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3714741498078457e-05,
+      "loss": 0.6428,
+      "step": 14500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2808353273874268e-05,
+      "loss": 0.622,
+      "step": 15000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1901965049670075e-05,
+      "loss": 0.6313,
+      "step": 15500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.0995576825465886e-05,
+      "loss": 0.6466,
+      "step": 16000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0089188601261694e-05,
+      "loss": 0.6203,
+      "step": 16500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.91828003770575e-05,
+      "loss": 0.6103,
+      "step": 17000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8276412152853312e-05,
+      "loss": 0.6085,
+      "step": 17500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.737002392864912e-05,
+      "loss": 0.6039,
+      "step": 18000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.646363570444493e-05,
+      "loss": 0.5586,
+      "step": 18500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5557247480240738e-05,
+      "loss": 0.4023,
+      "step": 19000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4650859256036547e-05,
+      "loss": 0.39,
+      "step": 19500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3744471031832355e-05,
+      "loss": 0.3777,
+      "step": 20000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2838082807628162e-05,
+      "loss": 0.3958,
+      "step": 20500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1931694583423973e-05,
+      "loss": 0.4169,
+      "step": 21000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1025306359219782e-05,
+      "loss": 0.3802,
+      "step": 21500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.011891813501559e-05,
+      "loss": 0.3793,
+      "step": 22000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.212529910811399e-06,
+      "loss": 0.4118,
+      "step": 22500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.306141686607208e-06,
+      "loss": 0.3776,
+      "step": 23000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.399753462403016e-06,
+      "loss": 0.3892,
+      "step": 23500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.493365238198826e-06,
+      "loss": 0.3759,
+      "step": 24000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.586977013994634e-06,
+      "loss": 0.3663,
+      "step": 24500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.680588789790443e-06,
+      "loss": 0.3836,
+      "step": 25000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.7742005655862523e-06,
+      "loss": 0.3731,
+      "step": 25500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8678123413820606e-06,
+      "loss": 0.3657,
+      "step": 26000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.96142411717787e-06,
+      "loss": 0.3635,
+      "step": 26500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0550358929736786e-06,
+      "loss": 0.3885,
+      "step": 27000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4864766876948735e-07,
+      "loss": 0.3573,
+      "step": 27500
+    },
+    {
+      "epoch": 3.0,
+      "step": 27582,
+      "total_flos": 2.836008195999293e+16,
+      "train_loss": 0.6582008492790169,
+      "train_runtime": 15084.1881,
+      "train_samples_per_second": 14.627,
+      "train_steps_per_second": 1.829
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 27582,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "total_flos": 2.836008195999293e+16,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5aa6359a4275e54941850b46e99f67e3b63904f5cac60221b891f449b5277bbf
+size 4728