Add collaterals

README.md

@@ -0,0 +1,67 @@
+BERT-base tuned for Squadv1.1 is pruned with movement pruning algorithm in hybrid fashion, i.e. 32x32 block for self-attention layers, per-dimension grain size for ffn layers.
+```
+  eval_exact_match = 78.5241
+  eval_f1          = 86.4138
+  eval_samples     =   10784
+``` 
+This model is a replication of [block pruning paper](https://arxiv.org/abs/2109.04838) with its open-sourced codebase (forked and modified). 
+To reproduce this model, pls follow [documentation here](https://github.com/vuiseng9/nn_pruning/blob/reproduce-evaluation/reproduce-eval/readme.md) until step 2.
+
+# Eval
+The model can be evaluated out-of-the-box with HF QA example. Note that only pruned self-attention heads are discarded where pruned ffn dimension are sparsified instead of removal. Verified in v4.13
+```bash
+export CUDA_VISIBLE_DEVICES=0
+
+OUTDIR=eval-bert-base-squadv1-block-pruning-hybrid
+WORKDIR=transformers/examples/pytorch/question-answering
+cd $WORKDIR
+
+nohup python run_qa.py  \
+    --model_name_or_path vuiseng9/bert-base-squadv1-block-pruning-hybrid  \
+    --dataset_name squad  \
+    --do_eval  \
+    --per_device_eval_batch_size 16  \
+    --max_seq_length 384  \
+    --doc_stride 128  \
+    --overwrite_output_dir \
+    --output_dir $OUTDIR 2>&1 | tee $OUTDIR/run.log &
+```
+
+If the intent is to observe inference acceleration, the pruned structure in the model must be "cropped"/discarded. Follow the custom setup below.
+```bash
+# OpenVINO/NNCF
+git clone https://github.com/vuiseng9/nncf && cd nncf
+git checkout tld-poc
+git reset --hard 1dec7afe7a4b567c059fcf287ea2c234980fded2
+python setup.py develop
+
+# Huggingface Transformers
+git clone https://github.com/vuiseng9/transformers && cd transformers
+git checkout tld-poc
+git reset --hard 10a1e29d84484e48fd106f58957d9ffc89dc43c5
+pip install -e .
+
+# Huggingface nn_pruning
+git clone https://github.com/vuiseng9/nn_pruning && cd nn_pruning
+git checkout reproduce-evaluation
+git reset --hard 2d4e196d694c465e43e5fbce6c3836d0a60e1446
+```
+Add ```--optimize_model_before_eval``` during evaluation.
+```bash
+export CUDA_VISIBLE_DEVICES=0
+
+OUTDIR=eval-bert-base-squadv1-block-pruning-hybrid-cropped
+WORKDIR=transformers/examples/pytorch/question-answering
+cd $WORKDIR
+
+nohup python run_qa.py  \
+    --model_name_or_path vuiseng9/bert-base-squadv1-block-pruning-hybrid  \
+    --dataset_name squad  \
+    --optimize_model_before_eval \
+    --do_eval  \
+    --per_device_eval_batch_size 16  \
+    --max_seq_length 384  \
+    --doc_stride 128  \
+    --overwrite_output_dir \
+    --output_dir $OUTDIR 2>&1 | tee $OUTDIR/run.log &
+```

config.json

@@ -0,0 +1,115 @@
+{
+  "_name_or_path": "bert-base-uncased",
+  "architectures": [
+    "BertForQuestionAnswering"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "classifier_dropout": null,
+  "gradient_checkpointing": false,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-12,
+  "max_position_embeddings": 512,
+  "model_type": "bert",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "pad_token_id": 0,
+  "position_embedding_type": "absolute",
+  "pruned_heads": {
+    "0": [
+      0,
+      2,
+      4,
+      5,
+      6,
+      7,
+      11
+    ],
+    "1": [
+      0,
+      2,
+      3,
+      5,
+      6,
+      7,
+      8
+    ],
+    "2": [
+      8,
+      4,
+      7
+    ],
+    "3": [
+      2,
+      4,
+      6
+    ],
+    "4": [
+      1,
+      2,
+      11
+    ],
+    "5": [
+      1,
+      2,
+      5,
+      6,
+      7,
+      11
+    ],
+    "6": [
+      0,
+      2,
+      3,
+      7,
+      10
+    ],
+    "7": [
+      1,
+      3,
+      6,
+      7,
+      11
+    ],
+    "8": [
+      0,
+      3,
+      4,
+      5,
+      8
+    ],
+    "9": [
+      1,
+      3,
+      4,
+      5,
+      7,
+      9,
+      10
+    ],
+    "10": [
+      1,
+      4,
+      5,
+      6,
+      7,
+      8
+    ],
+    "11": [
+      4,
+      5,
+      7,
+      8,
+      10,
+      11
+    ]
+  },
+  "torch_dtype": "float32",
+  "transformers_version": "4.10.3",
+  "type_vocab_size": 2,
+  "use_cache": true,
+  "vocab_size": 30522
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:895559c3ab4ac710f1747e5a3d2b5a45fa34edc8975bd9843f3d5960588f05b6
+size 386062513

special_tokens_map.json

@@ -0,0 +1 @@
+{"unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]"}

tokenizer_config.json

@@ -0,0 +1 @@
+{"do_lower_case": true, "unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]", "tokenize_chinese_chars": true, "strip_accents": null, "model_max_length": 512, "special_tokens_map_file": null, "name_or_path": "bert-base-uncased", "tokenizer_class": "BertTokenizer"}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f8baa4f88c06c0a3ab896b378e9ac834751b6b68df585f523ea3489516f5fecd
+size 2927