README.md

@@ -1,65 +1,64 @@
----
+---
-language: en
+language: en
-tags:
+tags:
-- tapex
+- tapex
-- table-question-answering
+license: apache-2.0
-license: mit
+---
----
+
-
+# TAPEX (base-sized model) 
-# TAPEX (base-sized model) 
+
-
+TAPEX was proposed in [TAPEX: Table Pre-training via Learning a Neural SQL Executor](https://arxiv.org/abs/2107.07653) by Qian Liu, Bei Chen, Jiaqi Guo, Morteza Ziyadi, Zeqi Lin, Weizhu Chen, Jian-Guang Lou. The original repo can be found [here](https://github.com/microsoft/Table-Pretraining).
-TAPEX was proposed in [TAPEX: Table Pre-training via Learning a Neural SQL Executor](https://arxiv.org/abs/2107.07653) by Qian Liu, Bei Chen, Jiaqi Guo, Morteza Ziyadi, Zeqi Lin, Weizhu Chen, Jian-Guang Lou. The original repo can be found [here](https://github.com/microsoft/Table-Pretraining).
+
-
+## Model description
-## Model description
+
-
+TAPEX (**Ta**ble **P**re-training via **Ex**ecution) is a conceptually simple and empirically powerful pre-training approach to empower existing models with *table reasoning* skills. TAPEX realizes table pre-training by learning a neural SQL executor over a synthetic corpus, which is obtained by automatically synthesizing executable SQL queries.
-TAPEX (**Ta**ble **P**re-training via **Ex**ecution) is a conceptually simple and empirically powerful pre-training approach to empower existing models with *table reasoning* skills. TAPEX realizes table pre-training by learning a neural SQL executor over a synthetic corpus, which is obtained by automatically synthesizing executable SQL queries.
+
-
+TAPEX is based on the BART architecture, the transformer encoder-encoder (seq2seq) model with a bidirectional (BERT-like) encoder and an autoregressive (GPT-like) decoder.
-TAPEX is based on the BART architecture, the transformer encoder-decoder (seq2seq) model with a bidirectional (BERT-like) encoder and an autoregressive (GPT-like) decoder.
+
-
+## Intended Uses
-## Intended Uses
+
-
+You can use the raw model for simulating neural SQL execution, i.e., employ TAPEX to execute a SQL query on a given table. However, the model is mostly meant to be fine-tuned on a supervised dataset. Currently TAPEX can be fine-tuned to tackle table question answering tasks and table fact verification tasks. See the [model hub](https://huggingface.co/models?search=tapex) to look for fine-tuned versions on a task that interests you.
-You can use the raw model for simulating neural SQL execution, i.e., employ TAPEX to execute a SQL query on a given table. However, the model is mostly meant to be fine-tuned on a supervised dataset. Currently TAPEX can be fine-tuned to tackle table question answering tasks and table fact verification tasks. See the [model hub](https://huggingface.co/models?search=tapex) to look for fine-tuned versions on a task that interests you.
+
-
+### How to Use
-### How to Use
+
-
+Here is how to use this model in transformers:
-Here is how to use this model in transformers:
+
-
+```python
-```python
+from transformers import TapexTokenizer, BartForConditionalGeneration
-from transformers import TapexTokenizer, BartForConditionalGeneration
+import pandas as pd
-import pandas as pd
+
-
+tokenizer = TapexTokenizer.from_pretrained("microsoft/tapex-base")
-tokenizer = TapexTokenizer.from_pretrained("microsoft/tapex-base")
+model = BartForConditionalGeneration.from_pretrained("microsoft/tapex-base")
-model = BartForConditionalGeneration.from_pretrained("microsoft/tapex-base")
+
-
+data = {
-data = {
+    "year": [1896, 1900, 1904, 2004, 2008, 2012],
-    "year": [1896, 1900, 1904, 2004, 2008, 2012],
+    "city": ["athens", "paris", "st. louis", "athens", "beijing", "london"]
-    "city": ["athens", "paris", "st. louis", "athens", "beijing", "london"]
+}
-}
+table = pd.DataFrame.from_dict(data)
-table = pd.DataFrame.from_dict(data)
+
-
+# tapex accepts uncased input since it is pre-trained on the uncased corpus
-# tapex accepts uncased input since it is pre-trained on the uncased corpus
+query = "select year where city = beijing"
-query = "select year where city = beijing"
+encoding = tokenizer(table=table, query=query, return_tensors="pt")
-encoding = tokenizer(table=table, query=query, return_tensors="pt")
+
-
+outputs = model.generate(**encoding)
-outputs = model.generate(**encoding)
+
-
+print(tokenizer.batch_decode(outputs, skip_special_tokens=True))
-print(tokenizer.batch_decode(outputs, skip_special_tokens=True))
+# ['2008']
-# ['2008']
+```
-```
+
-
+### How to Fine-tuning
-### How to Fine-tuning
+
-
+Please find the fine-tuning script [here](https://github.com/SivilTaram/transformers/tree/add_tapex_bis/examples/research_projects/tapex).
-Please find the fine-tuning script [here](https://github.com/SivilTaram/transformers/tree/add_tapex_bis/examples/research_projects/tapex).
+
-
+### BibTeX entry and citation info
-### BibTeX entry and citation info
+
-
+```bibtex
-```bibtex
+@inproceedings{
-@inproceedings{
+    liu2022tapex,
-    liu2022tapex,
+    title={{TAPEX}: Table Pre-training via Learning a Neural {SQL} Executor},
-    title={{TAPEX}: Table Pre-training via Learning a Neural {SQL} Executor},
+    author={Qian Liu and Bei Chen and Jiaqi Guo and Morteza Ziyadi and Zeqi Lin and Weizhu Chen and Jian-Guang Lou},
-    author={Qian Liu and Bei Chen and Jiaqi Guo and Morteza Ziyadi and Zeqi Lin and Weizhu Chen and Jian-Guang Lou},
+    booktitle={International Conference on Learning Representations},
-    booktitle={International Conference on Learning Representations},
+    year={2022},
-    year={2022},
+    url={https://openreview.net/forum?id=O50443AsCP}
-    url={https://openreview.net/forum?id=O50443AsCP}
+}
-}
 ```

config.json

@@ -1,39 +1,73 @@
-{
+{
-  "_name_or_path": "microsoft/tapex-base",
+  "_name_or_path": "bart-base",
-  "activation_dropout": 0.0,
+  "activation_dropout": 0.0,
-  "activation_function": "gelu",
+  "activation_function": "gelu",
-  "architectures": [
+  "add_bias_logits": false,
-    "BartForConditionalGeneration"
+  "add_final_layer_norm": false,
-  ],
+  "architectures": [
-  "attention_dropout": 0.1,
+    "BartModel"
-  "bos_token_id": 0,
+  ],
-  "classifier_dropout": 0.0,
+  "attention_dropout": 0.1,
-  "d_model": 768,
+  "bos_token_id": 0,
-  "decoder_attention_heads": 12,
+  "classifier_dropout": 0.0,
-  "decoder_ffn_dim": 3072,
+  "d_model": 768,
-  "decoder_layerdrop": 0.0,
+  "decoder_attention_heads": 12,
-  "decoder_layers": 6,
+  "decoder_ffn_dim": 3072,
-  "decoder_start_token_id": 2,
+  "decoder_layerdrop": 0.0,
-  "dropout": 0.1,
+  "decoder_layers": 6,
-  "encoder_attention_heads": 12,
+  "decoder_start_token_id": 2,
-  "encoder_ffn_dim": 3072,
+  "dropout": 0.1,
-  "encoder_layerdrop": 0.0,
+  "early_stopping": true,
-  "encoder_layers": 6,
+  "encoder_attention_heads": 12,
-  "eos_token_id": 2,
+  "encoder_ffn_dim": 3072,
-  "forced_bos_token_id": 0,
+  "encoder_layerdrop": 0.0,
-  "forced_eos_token_id": 2,
+  "encoder_layers": 6,
-  "init_std": 0.02,
+  "eos_token_id": 2,
-  "is_encoder_decoder": true,
+  "forced_eos_token_id": 2,
-  "max_position_embeddings": 1024,
+  "gradient_checkpointing": false,
-  "model_type": "bart",
+  "id2label": {
-  "normalize_before": false,
+    "0": "LABEL_0",
-  "normalize_embedding": true,
+    "1": "LABEL_1",
-  "num_beams": 4,
+    "2": "LABEL_2"
-  "num_hidden_layers": 6,
+  },
-  "pad_token_id": 1,
+  "init_std": 0.02,
-  "scale_embedding": false,
+  "is_encoder_decoder": true,
-  "torch_dtype": "float32",
+  "label2id": {
-  "transformers_version": "4.17.0.dev0",
+    "LABEL_0": 0,
-  "use_cache": true,
+    "LABEL_1": 1,
-  "vocab_size": 50265
+    "LABEL_2": 2
-}
+  },
+  "max_position_embeddings": 1024,
+  "model_type": "bart",
+  "no_repeat_ngram_size": 3,
+  "normalize_before": false,
+  "normalize_embedding": true,
+  "num_beams": 4,
+  "num_hidden_layers": 6,
+  "pad_token_id": 1,
+  "scale_embedding": false,
+  "task_specific_params": {
+    "summarization": {
+      "length_penalty": 1.0,
+      "max_length": 128,
+      "min_length": 12,
+      "num_beams": 4
+    },
+    "summarization_cnn": {
+      "length_penalty": 2.0,
+      "max_length": 142,
+      "min_length": 56,
+      "num_beams": 4
+    },
+    "summarization_xsum": {
+      "length_penalty": 1.0,
+      "max_length": 62,
+      "min_length": 11,
+      "num_beams": 6
+    }
+  },
+  "torch_dtype": "float32",
+  "transformers_version": "4.12.0.dev0",
+  "use_cache": true,
+  "vocab_size": 50265
+}

generation_config.json

@@ -1,11 +0,0 @@
-{
-  "_from_model_config": true,
-  "bos_token_id": 0,
-  "decoder_start_token_id": 2,
-  "eos_token_id": 2,
-  "forced_bos_token_id": 0,
-  "forced_eos_token_id": 2,
-  "num_beams": 4,
-  "pad_token_id": 1,
-  "transformers_version": "4.27.0.dev0"
-}

tokenizer_config.json

@@ -1 +0,0 @@
-{"do_lower_case": true, "errors": "replace", "bos_token": {"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "eos_token": {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "unk_token": {"content": "<unk>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "sep_token": {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "cls_token": {"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "pad_token": {"content": "<pad>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "mask_token": {"content": "<mask>", "single_word": false, "lstrip": true, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "add_prefix_space": false, "max_cell_length": 15, "model_max_length": 1024, "special_tokens_map_file": null, "name_or_path": "microsoft/tapex-base", "tokenizer_class": "TapexTokenizer"}