Add: add model weights and an example code.

Critique_NER/config.json

@@ -0,0 +1,58 @@
+{
+  "_name_or_path": "t5-large",
+  "architectures": [
+    "T5ForConditionalGeneration"
+  ],
+  "d_ff": 4096,
+  "d_kv": 64,
+  "d_model": 1024,
+  "decoder_start_token_id": 0,
+  "dropout_rate": 0.1,
+  "eos_token_id": 1,
+  "feed_forward_proj": "relu",
+  "initializer_factor": 1.0,
+  "is_encoder_decoder": true,
+  "layer_norm_epsilon": 1e-06,
+  "model_type": "t5",
+  "n_positions": 512,
+  "num_decoder_layers": 24,
+  "num_heads": 16,
+  "num_layers": 24,
+  "output_past": true,
+  "pad_token_id": 0,
+  "relative_attention_max_distance": 128,
+  "relative_attention_num_buckets": 32,
+  "task_specific_params": {
+    "summarization": {
+      "early_stopping": true,
+      "length_penalty": 2.0,
+      "max_length": 200,
+      "min_length": 30,
+      "no_repeat_ngram_size": 3,
+      "num_beams": 4,
+      "prefix": "summarize: "
+    },
+    "translation_en_to_de": {
+      "early_stopping": true,
+      "max_length": 300,
+      "num_beams": 4,
+      "prefix": "translate English to German: "
+    },
+    "translation_en_to_fr": {
+      "early_stopping": true,
+      "max_length": 300,
+      "num_beams": 4,
+      "prefix": "translate English to French: "
+    },
+    "translation_en_to_ro": {
+      "early_stopping": true,
+      "max_length": 300,
+      "num_beams": 4,
+      "prefix": "translate English to Romanian: "
+    }
+  },
+  "torch_dtype": "float32",
+  "transformers_version": "4.18.0",
+  "use_cache": true,
+  "vocab_size": 32128
+}

Critique_NER/pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c4ce87e207dea8a2cc2eb1efdfaec552f210d4e289bb43a136e1a6dffd8570b4
+size 2950844807

Critique_NUM/config.json

@@ -0,0 +1,58 @@
+{
+  "_name_or_path": "t5-large",
+  "architectures": [
+    "T5ForConditionalGeneration"
+  ],
+  "d_ff": 4096,
+  "d_kv": 64,
+  "d_model": 1024,
+  "decoder_start_token_id": 0,
+  "dropout_rate": 0.1,
+  "eos_token_id": 1,
+  "feed_forward_proj": "relu",
+  "initializer_factor": 1.0,
+  "is_encoder_decoder": true,
+  "layer_norm_epsilon": 1e-06,
+  "model_type": "t5",
+  "n_positions": 512,
+  "num_decoder_layers": 24,
+  "num_heads": 16,
+  "num_layers": 24,
+  "output_past": true,
+  "pad_token_id": 0,
+  "relative_attention_max_distance": 128,
+  "relative_attention_num_buckets": 32,
+  "task_specific_params": {
+    "summarization": {
+      "early_stopping": true,
+      "length_penalty": 2.0,
+      "max_length": 200,
+      "min_length": 30,
+      "no_repeat_ngram_size": 3,
+      "num_beams": 4,
+      "prefix": "summarize: "
+    },
+    "translation_en_to_de": {
+      "early_stopping": true,
+      "max_length": 300,
+      "num_beams": 4,
+      "prefix": "translate English to German: "
+    },
+    "translation_en_to_fr": {
+      "early_stopping": true,
+      "max_length": 300,
+      "num_beams": 4,
+      "prefix": "translate English to French: "
+    },
+    "translation_en_to_ro": {
+      "early_stopping": true,
+      "max_length": 300,
+      "num_beams": 4,
+      "prefix": "translate English to Romanian: "
+    }
+  },
+  "torch_dtype": "float32",
+  "transformers_version": "4.18.0",
+  "use_cache": true,
+  "vocab_size": 32128
+}

Critique_NUM/pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a1fa6ac29dd9298707e28d765d1f679bed36730dad4cbcef41ea43d47ab02d41
+size 2950844807

tag_critiques.py

@@ -0,0 +1,93 @@
+import sys
+import json
+from transformers import T5Tokenizer, T5ForConditionalGeneration
+import torch
+
+critic_model_num = "./Critique_NER"
+critic_model_ner = "./Critique_NUM"
+
+finitres = open(sys.argv[1], "r")
+
+initres = finitres.readlines()
+out_list = []
+
+for i in range(len(initres)):
+
+    res_record = json.loads(initres[i].strip())
+    out_text = ""
+    out_text +=  res_record["initial_response"]
+    out_text += " ||| Facts: " + res_record["facts"]
+
+    res_record["text"] = out_text
+    out_list.append(res_record)
+
+device = torch.device("cuda:0") if torch.cuda.is_available() else torch.device("cpu")
+# device = torch.device("cpu")
+
+# Prepare model
+tokenizer = T5Tokenizer.from_pretrained("t5-large")
+model = T5ForConditionalGeneration.from_pretrained(critic_model_num).to(device)
+
+
+task_prefix = "critique: "
+
+for i in range(len(out_list)):
+
+    sentences = [out_list[i]["text"]]
+    # print(sentences)
+
+    inputs = tokenizer([task_prefix + sentence for sentence in sentences], return_tensors="pt", padding=True).to(device)
+
+    output_sequences = model.generate(
+        input_ids=inputs["input_ids"],
+        attention_mask=inputs["attention_mask"],
+        do_sample=False,  # disable sampling to test if batching affects output
+    )
+
+    print(tokenizer.batch_decode(output_sequences, skip_special_tokens=True))
+    out_list[i]["critic_num"] = tokenizer.batch_decode(output_sequences, skip_special_tokens=True)[0]
+
+    del inputs
+    del output_sequences
+
+
+# Free gpu memory after using one of the T5 model
+del tokenizer
+del model
+torch.cuda.empty_cache()
+
+
+# Prepare model
+tokenizer = T5Tokenizer.from_pretrained("t5-large")
+model = T5ForConditionalGeneration.from_pretrained(critic_model_ner).to(device)
+
+task_prefix = "critique: "
+
+for i in range(len(out_list)):
+
+    sentences = [out_list[i]["text"]]
+
+    inputs = tokenizer([task_prefix + sentence for sentence in sentences], return_tensors="pt", padding=True).to(device)
+
+    output_sequences = model.generate(
+        input_ids=inputs["input_ids"],
+        attention_mask=inputs["attention_mask"],
+        do_sample=False,  # disable sampling to test if batching affects output
+    )
+
+    print(tokenizer.batch_decode(output_sequences, skip_special_tokens=True))
+    out_list[i]["critic_ner"] = tokenizer.batch_decode(output_sequences, skip_special_tokens=True)[0]
+
+    del inputs
+    del output_sequences
+
+
+# Free gpu memory after using one of the T5 model
+del tokenizer
+del model
+torch.cuda.empty_cache()
+
+
+fmerged_critics = open(sys.argv[2], "w")
+for i in out_list:
+    fmerged_critics.write(json.dumps(i) + "\n")