init commit

.gitignore

@@ -0,0 +1 @@
+__pycache__

UniPoll-t5/best_model/added_tokens.json

@@ -0,0 +1,6 @@
+{
+  "<c>": 50001,
+  "<choices>": 50003,
+  "<title>": 50002,
+  "digit": 50000
+}

UniPoll-t5/best_model/config.json

@@ -0,0 +1,32 @@
+{
+  "_name_or_path": "/root/Documents/CODES/UniPoll/checkpoints/imxly/t5-pegasus",
+  "architectures": [
+    "MT5ForConditionalGeneration"
+  ],
+  "d_ff": 2048,
+  "d_kv": 64,
+  "d_model": 768,
+  "decoder_start_token_id": 0,
+  "dense_act_fn": "gelu_new",
+  "dropout_rate": 0.1,
+  "eos_token_id": 1,
+  "feed_forward_proj": "gated-gelu",
+  "initializer_factor": 1.0,
+  "is_encoder_decoder": true,
+  "is_gated_act": true,
+  "layer_norm_epsilon": 1e-06,
+  "model_type": "mt5",
+  "num_decoder_layers": 12,
+  "num_heads": 12,
+  "num_layers": 12,
+  "output_past": true,
+  "pad_token_id": 0,
+  "relative_attention_max_distance": 128,
+  "relative_attention_num_buckets": 32,
+  "tie_word_embeddings": false,
+  "tokenizer_class": "T5Tokenizer",
+  "torch_dtype": "float32",
+  "transformers_version": "4.22.1",
+  "use_cache": true,
+  "vocab_size": 50004
+}

UniPoll-t5/best_model/pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1741c956f991a7ba239b38322657b7780961f9831914604a5adb75297f6f837e
+size 1100236621

UniPoll-t5/best_model/special_tokens_map.json

@@ -0,0 +1,7 @@
+{
+  "cls_token": "[CLS]",
+  "mask_token": "[MASK]",
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "unk_token": "[UNK]"
+}

UniPoll-t5/best_model/tokenizer_config.json

@@ -0,0 +1,15 @@
+{
+  "cls_token": "[CLS]",
+  "do_basic_tokenize": true,
+  "do_lower_case": true,
+  "mask_token": "[MASK]",
+  "name_or_path": "/root/Documents/CODES/UniPoll/checkpoints/imxly/t5-pegasus",
+  "never_split": null,
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "special_tokens_map_file": null,
+  "strip_accents": null,
+  "tokenize_chinese_chars": true,
+  "tokenizer_class": "T5PegasusTokenizer",
+  "unk_token": "[UNK]"
+}

UniPoll-t5/best_model/trainer_state.json

@@ -0,0 +1,1001 @@
+{
+  "best_metric": 49.33,
+  "best_model_checkpoint": "outputs/UniPoll-t5/checkpoint-60760",
+  "epoch": 10.0,
+  "global_step": 60760,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9753127057274523e-05,
+      "loss": 16.5215,
+      "step": 500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9506254114549046e-05,
+      "loss": 3.5706,
+      "step": 1000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9259381171823572e-05,
+      "loss": 2.9107,
+      "step": 1500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.901250822909809e-05,
+      "loss": 2.6763,
+      "step": 2000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.8765635286372614e-05,
+      "loss": 2.4391,
+      "step": 2500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.8518762343647136e-05,
+      "loss": 2.2868,
+      "step": 3000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.827188940092166e-05,
+      "loss": 2.2185,
+      "step": 3500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.802501645819618e-05,
+      "loss": 2.0841,
+      "step": 4000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.7778143515470708e-05,
+      "loss": 2.013,
+      "step": 4500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.7531270572745227e-05,
+      "loss": 1.9524,
+      "step": 5000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.728439763001975e-05,
+      "loss": 1.9129,
+      "step": 5500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7037524687294272e-05,
+      "loss": 1.8501,
+      "step": 6000
+    },
+    {
+      "epoch": 1.0,
+      "eval_choices_bleu1": 34.2743,
+      "eval_choices_bleu2": 27.071,
+      "eval_choices_bleu3": 23.8714,
+      "eval_choices_bleu4": 19.9582,
+      "eval_choices_rouge1": 43.056,
+      "eval_choices_rouge2": 30.797,
+      "eval_choices_rougeL": 40.79,
+      "eval_choices_rougeLsum": 30.362,
+      "eval_loss": 1.7218139171600342,
+      "eval_mean_rouge1": 44.783500000000004,
+      "eval_runtime": 613.1086,
+      "eval_samples_per_second": 3.303,
+      "eval_steps_per_second": 0.414,
+      "eval_title_bleu1": 39.1598,
+      "eval_title_bleu2": 24.8472,
+      "eval_title_bleu3": 17.1189,
+      "eval_title_bleu4": 11.4654,
+      "eval_title_rouge1": 46.511,
+      "eval_title_rouge2": 28.336,
+      "eval_title_rougeL": 43.814,
+      "eval_title_rougeLsum": 28.185,
+      "step": 6076
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6790651744568795e-05,
+      "loss": 1.7659,
+      "step": 6500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.654377880184332e-05,
+      "loss": 1.7064,
+      "step": 7000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.6296905859117843e-05,
+      "loss": 1.6739,
+      "step": 7500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6050032916392366e-05,
+      "loss": 1.6589,
+      "step": 8000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.5803159973666885e-05,
+      "loss": 1.6193,
+      "step": 8500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5556287030941408e-05,
+      "loss": 1.6175,
+      "step": 9000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.530941408821593e-05,
+      "loss": 1.6042,
+      "step": 9500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5062541145490456e-05,
+      "loss": 1.5438,
+      "step": 10000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.481566820276498e-05,
+      "loss": 1.5275,
+      "step": 10500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.45687952600395e-05,
+      "loss": 1.5144,
+      "step": 11000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.432192231731402e-05,
+      "loss": 1.4911,
+      "step": 11500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.4075049374588544e-05,
+      "loss": 1.4301,
+      "step": 12000
+    },
+    {
+      "epoch": 2.0,
+      "eval_choices_bleu1": 36.2304,
+      "eval_choices_bleu2": 29.2436,
+      "eval_choices_bleu3": 26.0301,
+      "eval_choices_bleu4": 21.8164,
+      "eval_choices_rouge1": 45.164,
+      "eval_choices_rouge2": 33.104,
+      "eval_choices_rougeL": 42.835,
+      "eval_choices_rougeLsum": 32.498,
+      "eval_loss": 1.6042917966842651,
+      "eval_mean_rouge1": 47.2485,
+      "eval_runtime": 613.8133,
+      "eval_samples_per_second": 3.299,
+      "eval_steps_per_second": 0.414,
+      "eval_title_bleu1": 41.7754,
+      "eval_title_bleu2": 27.1102,
+      "eval_title_bleu3": 18.344,
+      "eval_title_bleu4": 12.3234,
+      "eval_title_rouge1": 49.333,
+      "eval_title_rouge2": 30.676,
+      "eval_title_rougeL": 46.561,
+      "eval_title_rougeLsum": 30.748,
+      "step": 12152
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.3828176431863066e-05,
+      "loss": 1.4002,
+      "step": 12500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3581303489137592e-05,
+      "loss": 1.3791,
+      "step": 13000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.3334430546412115e-05,
+      "loss": 1.3332,
+      "step": 13500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3087557603686637e-05,
+      "loss": 1.3577,
+      "step": 14000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.284068466096116e-05,
+      "loss": 1.3483,
+      "step": 14500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.259381171823568e-05,
+      "loss": 1.3584,
+      "step": 15000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.2346938775510205e-05,
+      "loss": 1.3186,
+      "step": 15500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.2100065832784728e-05,
+      "loss": 1.3283,
+      "step": 16000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.185319289005925e-05,
+      "loss": 1.2867,
+      "step": 16500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1606319947333773e-05,
+      "loss": 1.2766,
+      "step": 17000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1359447004608296e-05,
+      "loss": 1.3025,
+      "step": 17500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.111257406188282e-05,
+      "loss": 1.2736,
+      "step": 18000
+    },
+    {
+      "epoch": 3.0,
+      "eval_choices_bleu1": 36.9274,
+      "eval_choices_bleu2": 30.039,
+      "eval_choices_bleu3": 26.4107,
+      "eval_choices_bleu4": 22.0831,
+      "eval_choices_rouge1": 46.229,
+      "eval_choices_rouge2": 33.898,
+      "eval_choices_rougeL": 43.746,
+      "eval_choices_rougeLsum": 33.141,
+      "eval_loss": 1.5690504312515259,
+      "eval_mean_rouge1": 48.167500000000004,
+      "eval_runtime": 614.3575,
+      "eval_samples_per_second": 3.296,
+      "eval_steps_per_second": 0.413,
+      "eval_title_bleu1": 42.3694,
+      "eval_title_bleu2": 28.0839,
+      "eval_title_bleu3": 19.3916,
+      "eval_title_bleu4": 13.1408,
+      "eval_title_rouge1": 50.106,
+      "eval_title_rouge2": 31.608,
+      "eval_title_rougeL": 47.1,
+      "eval_title_rougeLsum": 31.399,
+      "step": 18228
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.086570111915734e-05,
+      "loss": 1.2176,
+      "step": 18500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.0618828176431864e-05,
+      "loss": 1.2037,
+      "step": 19000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.0371955233706386e-05,
+      "loss": 1.2104,
+      "step": 19500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.012508229098091e-05,
+      "loss": 1.1522,
+      "step": 20000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.987820934825543e-05,
+      "loss": 1.1969,
+      "step": 20500
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.9631336405529957e-05,
+      "loss": 1.1861,
+      "step": 21000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.9384463462804477e-05,
+      "loss": 1.1536,
+      "step": 21500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.9137590520079e-05,
+      "loss": 1.1394,
+      "step": 22000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.8890717577353522e-05,
+      "loss": 1.1229,
+      "step": 22500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.8643844634628044e-05,
+      "loss": 1.1233,
+      "step": 23000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8396971691902567e-05,
+      "loss": 1.1265,
+      "step": 23500
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.8150098749177093e-05,
+      "loss": 1.1438,
+      "step": 24000
+    },
+    {
+      "epoch": 4.0,
+      "eval_choices_bleu1": 38.2715,
+      "eval_choices_bleu2": 30.7803,
+      "eval_choices_bleu3": 26.9954,
+      "eval_choices_bleu4": 22.6626,
+      "eval_choices_rouge1": 46.767,
+      "eval_choices_rouge2": 34.301,
+      "eval_choices_rougeL": 44.062,
+      "eval_choices_rougeLsum": 33.605,
+      "eval_loss": 1.5627468824386597,
+      "eval_mean_rouge1": 48.6515,
+      "eval_runtime": 625.5501,
+      "eval_samples_per_second": 3.237,
+      "eval_steps_per_second": 0.406,
+      "eval_title_bleu1": 42.7056,
+      "eval_title_bleu2": 28.0993,
+      "eval_title_bleu3": 19.498,
+      "eval_title_bleu4": 12.7399,
+      "eval_title_rouge1": 50.536,
+      "eval_title_rouge2": 31.852,
+      "eval_title_rougeL": 47.453,
+      "eval_title_rougeLsum": 31.678,
+      "step": 24304
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.7903225806451616e-05,
+      "loss": 1.0962,
+      "step": 24500
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.7656352863726135e-05,
+      "loss": 1.0664,
+      "step": 25000
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.7409479921000658e-05,
+      "loss": 1.0874,
+      "step": 25500
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.716260697827518e-05,
+      "loss": 1.0608,
+      "step": 26000
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.6915734035549703e-05,
+      "loss": 1.0367,
+      "step": 26500
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.666886109282423e-05,
+      "loss": 1.0397,
+      "step": 27000
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.642198815009875e-05,
+      "loss": 1.0378,
+      "step": 27500
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.617511520737327e-05,
+      "loss": 1.0143,
+      "step": 28000
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.5928242264647793e-05,
+      "loss": 1.0186,
+      "step": 28500
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5681369321922316e-05,
+      "loss": 1.0243,
+      "step": 29000
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.5434496379196842e-05,
+      "loss": 1.038,
+      "step": 29500
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5187623436471364e-05,
+      "loss": 1.0026,
+      "step": 30000
+    },
+    {
+      "epoch": 5.0,
+      "eval_choices_bleu1": 37.513,
+      "eval_choices_bleu2": 30.5119,
+      "eval_choices_bleu3": 26.8236,
+      "eval_choices_bleu4": 22.3797,
+      "eval_choices_rouge1": 46.999,
+      "eval_choices_rouge2": 34.451,
+      "eval_choices_rougeL": 44.427,
+      "eval_choices_rougeLsum": 33.652,
+      "eval_loss": 1.5795127153396606,
+      "eval_mean_rouge1": 48.658500000000004,
+      "eval_runtime": 619.9047,
+      "eval_samples_per_second": 3.267,
+      "eval_steps_per_second": 0.41,
+      "eval_title_bleu1": 42.3561,
+      "eval_title_bleu2": 27.804,
+      "eval_title_bleu3": 19.0851,
+      "eval_title_bleu4": 12.6803,
+      "eval_title_rouge1": 50.318,
+      "eval_title_rouge2": 31.494,
+      "eval_title_rougeL": 47.297,
+      "eval_title_rougeLsum": 31.461,
+      "step": 30380
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4940750493745885e-05,
+      "loss": 1.0107,
+      "step": 30500
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.4693877551020408e-05,
+      "loss": 0.9664,
+      "step": 31000
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.4447004608294932e-05,
+      "loss": 0.9342,
+      "step": 31500
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.4200131665569453e-05,
+      "loss": 0.9632,
+      "step": 32000
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.3953258722843976e-05,
+      "loss": 0.9448,
+      "step": 32500
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.37063857801185e-05,
+      "loss": 0.9669,
+      "step": 33000
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.3459512837393023e-05,
+      "loss": 0.9361,
+      "step": 33500
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.3212639894667544e-05,
+      "loss": 0.9445,
+      "step": 34000
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.2965766951942068e-05,
+      "loss": 0.9515,
+      "step": 34500
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.271889400921659e-05,
+      "loss": 0.9488,
+      "step": 35000
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.2472021066491112e-05,
+      "loss": 0.9145,
+      "step": 35500
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.2225148123765636e-05,
+      "loss": 0.9493,
+      "step": 36000
+    },
+    {
+      "epoch": 6.0,
+      "eval_choices_bleu1": 38.7421,
+      "eval_choices_bleu2": 31.1954,
+      "eval_choices_bleu3": 27.0903,
+      "eval_choices_bleu4": 22.4639,
+      "eval_choices_rouge1": 47.617,
+      "eval_choices_rouge2": 34.86,
+      "eval_choices_rougeL": 44.679,
+      "eval_choices_rougeLsum": 34.003,
+      "eval_loss": 1.5864702463150024,
+      "eval_mean_rouge1": 48.966499999999996,
+      "eval_runtime": 615.3469,
+      "eval_samples_per_second": 3.291,
+      "eval_steps_per_second": 0.413,
+      "eval_title_bleu1": 42.4031,
+      "eval_title_bleu2": 28.2011,
+      "eval_title_bleu3": 19.3663,
+      "eval_title_bleu4": 12.8818,
+      "eval_title_rouge1": 50.316,
+      "eval_title_rouge2": 31.841,
+      "eval_title_rougeL": 47.179,
+      "eval_title_rougeLsum": 31.509,
+      "step": 36456
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.1978275181040158e-05,
+      "loss": 0.9359,
+      "step": 36500
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.1731402238314681e-05,
+      "loss": 0.9096,
+      "step": 37000
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.1484529295589204e-05,
+      "loss": 0.9069,
+      "step": 37500
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.1237656352863726e-05,
+      "loss": 0.89,
+      "step": 38000
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.099078341013825e-05,
+      "loss": 0.8761,
+      "step": 38500
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.0743910467412772e-05,
+      "loss": 0.8919,
+      "step": 39000
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.0497037524687294e-05,
+      "loss": 0.8869,
+      "step": 39500
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.0250164581961818e-05,
+      "loss": 0.8912,
+      "step": 40000
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.000329163923634e-05,
+      "loss": 0.8885,
+      "step": 40500
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 9.756418696510862e-06,
+      "loss": 0.8691,
+      "step": 41000
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 9.509545753785386e-06,
+      "loss": 0.8882,
+      "step": 41500
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 9.262672811059907e-06,
+      "loss": 0.8822,
+      "step": 42000
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.01579986833443e-06,
+      "loss": 0.8622,
+      "step": 42500
+    },
+    {
+      "epoch": 7.0,
+      "eval_choices_bleu1": 38.5191,
+      "eval_choices_bleu2": 31.1743,
+      "eval_choices_bleu3": 27.2188,
+      "eval_choices_bleu4": 22.5257,
+      "eval_choices_rouge1": 47.345,
+      "eval_choices_rouge2": 34.733,
+      "eval_choices_rougeL": 44.637,
+      "eval_choices_rougeLsum": 33.956,
+      "eval_loss": 1.6173765659332275,
+      "eval_mean_rouge1": 49.0085,
+      "eval_runtime": 624.2904,
+      "eval_samples_per_second": 3.244,
+      "eval_steps_per_second": 0.407,
+      "eval_title_bleu1": 42.7395,
+      "eval_title_bleu2": 28.2813,
+      "eval_title_bleu3": 19.4836,
+      "eval_title_bleu4": 12.7587,
+      "eval_title_rouge1": 50.672,
+      "eval_title_rouge2": 32.013,
+      "eval_title_rougeL": 47.817,
+      "eval_title_rougeLsum": 31.899,
+      "step": 42532
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 8.768926925608954e-06,
+      "loss": 0.8468,
+      "step": 43000
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.522053982883475e-06,
+      "loss": 0.8209,
+      "step": 43500
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.275181040158e-06,
+      "loss": 0.8379,
+      "step": 44000
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.028308097432522e-06,
+      "loss": 0.8544,
+      "step": 44500
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 7.781435154707045e-06,
+      "loss": 0.8394,
+      "step": 45000
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 7.534562211981568e-06,
+      "loss": 0.8285,
+      "step": 45500
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.28768926925609e-06,
+      "loss": 0.8333,
+      "step": 46000
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.0408163265306125e-06,
+      "loss": 0.8359,
+      "step": 46500
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.793943383805135e-06,
+      "loss": 0.8176,
+      "step": 47000
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.547070441079658e-06,
+      "loss": 0.842,
+      "step": 47500
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.30019749835418e-06,
+      "loss": 0.8351,
+      "step": 48000
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.053324555628703e-06,
+      "loss": 0.8367,
+      "step": 48500
+    },
+    {
+      "epoch": 8.0,
+      "eval_choices_bleu1": 38.7622,
+      "eval_choices_bleu2": 31.0867,
+      "eval_choices_bleu3": 26.9654,
+      "eval_choices_bleu4": 22.4556,
+      "eval_choices_rouge1": 47.531,
+      "eval_choices_rouge2": 34.756,
+      "eval_choices_rougeL": 44.649,
+      "eval_choices_rougeLsum": 33.959,
+      "eval_loss": 1.6293848752975464,
+      "eval_mean_rouge1": 49.2605,
+      "eval_runtime": 622.1469,
+      "eval_samples_per_second": 3.255,
+      "eval_steps_per_second": 0.408,
+      "eval_title_bleu1": 43.0901,
+      "eval_title_bleu2": 28.6471,
+      "eval_title_bleu3": 19.7616,
+      "eval_title_bleu4": 12.9726,
+      "eval_title_rouge1": 50.99,
+      "eval_title_rouge2": 32.44,
+      "eval_title_rougeL": 47.948,
+      "eval_title_rougeLsum": 32.24,
+      "step": 48608
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.8064516129032256e-06,
+      "loss": 0.8004,
+      "step": 49000
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.559578670177749e-06,
+      "loss": 0.789,
+      "step": 49500
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.312705727452272e-06,
+      "loss": 0.7843,
+      "step": 50000
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.065832784726793e-06,
+      "loss": 0.8102,
+      "step": 50500
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.818959842001317e-06,
+      "loss": 0.7979,
+      "step": 51000
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.5720868992758395e-06,
+      "loss": 0.8262,
+      "step": 51500
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.325213956550362e-06,
+      "loss": 0.8149,
+      "step": 52000
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.078341013824885e-06,
+      "loss": 0.8178,
+      "step": 52500
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.831468071099407e-06,
+      "loss": 0.7796,
+      "step": 53000
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.5845951283739304e-06,
+      "loss": 0.8036,
+      "step": 53500
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.337722185648453e-06,
+      "loss": 0.8169,
+      "step": 54000
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.0908492429229756e-06,
+      "loss": 0.8138,
+      "step": 54500
+    },
+    {
+      "epoch": 9.0,
+      "eval_choices_bleu1": 38.8037,
+      "eval_choices_bleu2": 31.1809,
+      "eval_choices_bleu3": 27.0585,
+      "eval_choices_bleu4": 22.3774,
+      "eval_choices_rouge1": 47.629,
+      "eval_choices_rouge2": 34.778,
+      "eval_choices_rougeL": 44.722,
+      "eval_choices_rougeLsum": 33.944,
+      "eval_loss": 1.6443856954574585,
+      "eval_mean_rouge1": 49.32,
+      "eval_runtime": 624.8802,
+      "eval_samples_per_second": 3.241,
+      "eval_steps_per_second": 0.406,
+      "eval_title_bleu1": 43.0453,
+      "eval_title_bleu2": 28.6763,
+      "eval_title_bleu3": 19.7491,
+      "eval_title_bleu4": 13.004,
+      "eval_title_rouge1": 51.011,
+      "eval_title_rouge2": 32.398,
+      "eval_title_rougeL": 47.974,
+      "eval_title_rougeLsum": 32.251,
+      "step": 54684
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 2.8439763001974987e-06,
+      "loss": 0.7764,
+      "step": 55000
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.5971033574720213e-06,
+      "loss": 0.7939,
+      "step": 55500
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.3502304147465435e-06,
+      "loss": 0.8069,
+      "step": 56000
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.1033574720210665e-06,
+      "loss": 0.7649,
+      "step": 56500
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.8564845292955893e-06,
+      "loss": 0.7977,
+      "step": 57000
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.609611586570112e-06,
+      "loss": 0.803,
+      "step": 57500
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.3627386438446346e-06,
+      "loss": 0.7813,
+      "step": 58000
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.1158657011191574e-06,
+      "loss": 0.7645,
+      "step": 58500
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 8.689927583936801e-07,
+      "loss": 0.7865,
+      "step": 59000
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.221198156682028e-07,
+      "loss": 0.7818,
+      "step": 59500
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.752468729427255e-07,
+      "loss": 0.7572,
+      "step": 60000
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.283739302172482e-07,
+      "loss": 0.7788,
+      "step": 60500
+    },
+    {
+      "epoch": 10.0,
+      "eval_choices_bleu1": 38.8515,
+      "eval_choices_bleu2": 31.2992,
+      "eval_choices_bleu3": 27.2073,
+      "eval_choices_bleu4": 22.4633,
+      "eval_choices_rouge1": 47.688,
+      "eval_choices_rouge2": 34.913,
+      "eval_choices_rougeL": 44.77,
+      "eval_choices_rougeLsum": 34.038,
+      "eval_loss": 1.6531261205673218,
+      "eval_mean_rouge1": 49.33,
+      "eval_runtime": 620.225,
+      "eval_samples_per_second": 3.265,
+      "eval_steps_per_second": 0.41,
+      "eval_title_bleu1": 43.0602,
+      "eval_title_bleu2": 28.6495,
+      "eval_title_bleu3": 19.8146,
+      "eval_title_bleu4": 13.144,
+      "eval_title_rouge1": 50.972,
+      "eval_title_rouge2": 32.375,
+      "eval_title_rougeL": 47.919,
+      "eval_title_rougeLsum": 32.201,
+      "step": 60760
+    },
+    {
+      "epoch": 10.0,
+      "step": 60760,
+      "total_flos": 1.262557342857001e+17,
+      "train_loss": 1.288209796626761,
+      "train_runtime": 17551.2786,
+      "train_samples_per_second": 27.692,
+      "train_steps_per_second": 3.462
+    }
+  ],
+  "max_steps": 60760,
+  "num_train_epochs": 10,
+  "total_flos": 1.262557342857001e+17,
+  "trial_name": null,
+  "trial_params": null
+}

UniPoll-t5/best_model/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:53d331e9447f18c1699297a18369f397bdf5607c47ff32a23a2d6d9c4d795179
+size 3503

app.py

@@ -0,0 +1,170 @@
+import gc
+import sys
+import torch
+import argparse
+import gradio as gr
+from typing import List, Tuple
+from transformers import AutoConfig
+from transformers.models.mt5.modeling_mt5 import MT5ForConditionalGeneration
+from utils import T5PegasusTokenizer
+
+try:
+    from loguru import logger as logging
+    logging.add(sys.stderr, filter="my_module")
+except ImportError:
+    import logging
+
+import time
+class TimerDecorator:
+    def __init__(self, func) :
+        self.func = func
+
+    def __call__(self, *args, **kwargs) :
+        start_time = time.time()
+        result = self.func(*args, **kwargs)
+        end_time = time.time()
+        t = end_time - start_time
+        logging.info(f"Function `{self.func.__name__}` took {round(t, 2)} s to run.")
+        return result
+
+@TimerDecorator
+def load_model(model_path, device="cpu"):
+    logging.info(f"Loading model from {model_path}")
+    config = AutoConfig.from_pretrained(model_path)
+    tokenizer = T5PegasusTokenizer.from_pretrained(model_path)
+    model = MT5ForConditionalGeneration.from_pretrained(model_path, config=config)
+    if device != "cpu":
+        model.to(device)
+    logging.info("Done.")
+    return model, tokenizer
+
+def wrap_prompt(
+        post, comments, 
+        prompt="生成 <title> 和 <choices>: [SEP] {post} [SEP] {comments}"
+    ):
+    if not comments or comments == "":
+        logging.info("No comments input, comments will be ignored.")
+        prompt = prompt.replace(" [SEP] {comments}", "")
+        prompt = prompt.format(post=post)
+    else:
+        prompt = prompt.format(post=post, comments=comments)
+    logging.info(f"Wrapped prompt: {prompt}")
+    return prompt
+
+@TimerDecorator
+def generate(query, model, tokenizer, num_beams=4,  device="cpu"):
+    logging.info("Generating output...")
+    tokens = tokenizer(query, return_tensors="pt")["input_ids"]
+    if device != "cpu":
+        tokens = tokens.to(device)
+    output = model.generate(tokens, num_beams=num_beams, max_length=100)
+    output_text = tokenizer.batch_decode(output, skip_special_tokens=True)[0]
+    logging.info("Done.")
+    return output_text
+    
+def post_process(raw_output: str) -> Tuple[str, str]:
+    def same_title_choices(raw_output):
+        # return the same raw output as title and choices 
+        # if no <title> or <choices> in raw_output
+        raw_output = raw_output.replace("<title>", "")
+        raw_output = raw_output.replace("<choices>", "")
+        return raw_output.strip(), [raw_output.strip()]
+    
+    def split_choices(choices_str: str) -> List[str]:
+        choices = choices_str.split("<c>")
+        choices = [choice.strip() for choice in choices]
+        return choices
+
+    # extract title and choices from raw_output
+    # e.g. raw_output = "<title> 你 觉得 线 上 复试 公平 吗 <choices> 公平 <c> 不 公平"
+    if "<title>" in raw_output and "<choices>" in raw_output:
+        index1 = raw_output.index("<title>")
+        index2 = raw_output.index("<choices>")
+        if index1 > index2:
+            logging.debug(f"idx1>idx2, same title and choices will be used.\nraw_output: {raw_output}")
+            return same_title_choices(raw_output)
+        title = raw_output[index1+7: index2].strip()    # "你 觉得 线 上 复试 公平 吗"
+        choices_str = raw_output[index2+9:].strip()     # "公平 <c> 不 公平"
+        choices = split_choices(choices_str)            # ["公平", "不 公平"]
+    else:        
+        logging.debug(f"missing title/choices, same title and choices will be used.\nraw_output: {raw_output}")
+        title, choices = same_title_choices(raw_output)
+
+    def remove_blank(string):
+        return string.replace(" ", "")
+    
+    title = remove_blank(title)
+    choices = [remove_blank(choice) for choice in choices]
+    return title, choices
+
+def parse_args():
+    parser = argparse.ArgumentParser(description="Demo")
+    parser.add_argument("--model_path", type=str, default="./UniPoll-t5/best_model", help="path to the model.")
+    parser.add_argument("--device", type=str, default="cpu", help="specify the device to load the model, e.g. 'cpu', 'cuda:0'.")
+    parser.add_argument(
+        "--options",
+        nargs="+",
+        help="override some settings in the used config, the key-value pair "
+        "in xxx=yyy format will be merged into config file (deprecate), "
+        "change to --cfg-options instead.",
+    )
+    args = parser.parse_args()
+    return args
+
+
+if __name__ == "__main__":    
+    args = parse_args()
+    
+    logging.info('Initializing Model...')
+    # prepare the model
+    model, tokenizer = load_model(args.model_path, args.device)
+
+    def submit(post, comments, num_beams):
+        try:
+            logging.info("Received post input: {}".format(post))
+            if comments:
+                logging.info("Received comments input: {}".format(comments))
+
+            query = wrap_prompt(post, comments)
+            raw_output = generate(
+                query, model, tokenizer, num_beams, args.device)
+            title, choices = post_process(raw_output)         # post process
+            logging.info(f"Raw output: {raw_output}")
+            logging.info(f"Processed title: {title}")
+            logging.info(f"Processed choices: {choices}")
+            # return title, choices, raw_output
+            return title, choices
+        except Exception as e:
+            return "An error occurred: {}".format(str(e)), "An error occurred: {}".format(str(e))
+        finally:
+            gc.collect()
+            torch.cuda.empty_cache()
+
+    examples = [
+        ["#哪吒，大鱼海棠重映#动画电影《哪吒之魔童降世》、《大鱼海棠》，以及雷佳音、佟丽娅主演的 《超时空同居》确定将重映。据最新数据显示，3月24日全国复工影院495家，复工率4.36%，单日票房2.7万元。", "我在人间贩卖黄昏，只为收集世间温柔，去见你。谢谢你的分享，来看看你。我的微博，随时恭候你的到..."],
+        ["#线上复试是否能保障公平＃ 高考延期惹的祸，考研线上复试，那还能保证公平吗？", "这个世界上本来就没有绝对的公平。你可以说一个倒数第一考了第一，但考上了他也还是啥都不会。也可以说他会利用一切机会达到目的，反正结果就是人家考的好，你还找不出来证据。线上考试，平时考倒数的人进了年级前十。平时考试有水分，线上之后，那不就是在水里考？"],
+        ["#断亲现象为何如此流行#？所谓“断亲”指的是当代年轻人懒于、疏于、不屑于跟亲戚交往、联系、互动，日常音信全无，哪怕在逢年过节期间，宁可独来独往，也不愿意走亲戚，甚至将此作为一种时尚生活方式来推崇。", ""]
+    ]
+    
+    description = """This is the demo of UniPoll. Please input post and comments. <div style='display:flex; gap: 0.25rem; '><a href='https://uni-poll.github.io'><img src='https://img.shields.io/badge/Project-Page-Green'></a><a href='https://github.com/X1AOX1A/UniPoll'><img src='https://img.shields.io/badge/Github-Code-blue'></a><a href='https://arxiv.org/abs/2306.06851'><img src='https://img.shields.io/badge/Paper-PDF-red'></a></div>
+"""
+
+    demo = gr.Interface(
+        fn=submit,
+        inputs=[gr.Textbox(lines=1, label="Social Media Post", placeholder="Input post here..."),
+                gr.Textbox(lines=1, label="Social Media Comments (Optional)", placeholder="Input comments here..."),
+                gr.Number(value=4, label="Number of Beams", precision=0),
+            ],                
+        outputs=[gr.Textbox(lines=1, label="Generated Poll Question", placeholder="Generated poll question will be shown here"),
+                 gr.Textbox(lines=1, label="Generated Poll Choices", placeholder="Generated poll choices will be shown here"),
+           ], # question, choices
+        title="Demo of UniPoll",
+        description=description,
+        allow_flagging="never",
+        examples=examples,
+    )
+    
+    demo.queue(max_size=10)
+    demo.launch(share=True, show_error=True)
+
+# python app.py --model_path "./UniPoll-t5/best_model" --device "cpu"

requirements.txt

@@ -0,0 +1,410 @@
+absl-py==1.4.0
+aiofiles==23.2.1
+alabaster @ file:///home/ktietz/src/ci/alabaster_1611921544520/work
+altair==5.0.1
+anaconda-client==1.11.0
+anaconda-navigator==2.3.1
+anaconda-project @ file:///opt/conda/conda-bld/anaconda-project_1660339890420/work
+annotated-types==0.5.0
+antlr4-python3-runtime==4.8
+anyio==3.7.1
+appdirs==1.4.4
+argon2-cffi @ file:///opt/conda/conda-bld/argon2-cffi_1645000214183/work
+argon2-cffi-bindings @ file:///tmp/build/80754af9/argon2-cffi-bindings_1644569679365/work
+arrow @ file:///opt/conda/conda-bld/arrow_1649166651673/work
+astroid @ file:///tmp/abs_e5wkt48jiz/croots/recipe/astroid_1659023120113/work
+astropy @ file:///opt/conda/conda-bld/astropy_1657786094003/work
+astunparse==1.6.3
+atomicwrites==1.4.0
+attrs @ file:///opt/conda/conda-bld/attrs_1642510447205/work
+Automat @ file:///tmp/build/80754af9/automat_1600298431173/work
+autopep8 @ file:///opt/conda/conda-bld/autopep8_1650463822033/work
+Babel @ file:///tmp/build/80754af9/babel_1620871417480/work
+backcall @ file:///home/ktietz/src/ci/backcall_1611930011877/work
+backports.functools-lru-cache @ file:///tmp/build/80754af9/backports.functools_lru_cache_1618170165463/work
+backports.tempfile @ file:///home/linux1/recipes/ci/backports.tempfile_1610991236607/work
+backports.weakref==1.0.post1
+bcrypt @ file:///tmp/abs_6fpe92qzzo/croots/recipe/bcrypt_1659554336150/work
+beautifulsoup4 @ file:///opt/conda/conda-bld/beautifulsoup4_1650462163268/work
+binaryornot @ file:///tmp/build/80754af9/binaryornot_1617751525010/work
+bitarray @ file:///opt/conda/conda-bld/bitarray_1657739645104/work
+bkcharts==0.2
+black @ file:///opt/conda/conda-bld/black_1660237809219/work
+bleach @ file:///opt/conda/conda-bld/bleach_1641577558959/work
+blinker==1.6.2
+bokeh @ file:///tmp/abs_34854e1f-d7d3-4f22-85d9-1075588e4ecdga64o0qg/croots/recipe/bokeh_1658136654619/work
+boto3 @ file:///tmp/abs_ae3c72db-af47-4298-baea-7270430e2c96scbpg1_h/croots/recipe/boto3_1657820109150/work
+botocore @ file:///opt/conda/conda-bld/botocore_1657739486257/work
+Bottleneck @ file:///opt/conda/conda-bld/bottleneck_1657175564434/work
+brotlipy==0.7.0
+cachetools==5.3.0
+certifi @ file:///opt/conda/conda-bld/certifi_1663615672595/work/certifi
+cffi @ file:///tmp/abs_98z5h56wf8/croots/recipe/cffi_1659598650955/work
+chardet @ file:///tmp/build/80754af9/chardet_1607706775000/work
+charset-normalizer @ file:///tmp/build/80754af9/charset-normalizer_1630003229654/work
+click @ file:///tmp/build/80754af9/click_1646056590078/work
+cloudpickle @ file:///tmp/build/80754af9/cloudpickle_1632508026186/work
+clyent==1.2.2
+cmake==3.26.3
+colorama @ file:///opt/conda/conda-bld/colorama_1657009087971/work
+colorcet @ file:///tmp/build/80754af9/colorcet_1651851439427/work
+conda==23.1.0
+conda-build==3.22.0
+conda-content-trust @ file:///tmp/abs_5952f1c8-355c-4855-ad2e-538535021ba5h26t22e5/croots/recipe/conda-content-trust_1658126371814/work
+conda-pack @ file:///tmp/build/80754af9/conda-pack_1611163042455/work
+conda-package-handling @ file:///opt/conda/conda-bld/conda-package-handling_1663598473529/work
+conda-repo-cli==1.0.20
+conda-token @ file:///Users/paulyim/miniconda3/envs/c3i/conda-bld/conda-token_1662660369760/work
+conda-verify==3.4.2
+constantly==15.1.0
+cookiecutter @ file:///opt/conda/conda-bld/cookiecutter_1649151442564/work
+cryptography @ file:///tmp/build/80754af9/cryptography_1652101588893/work
+cssselect==1.1.0
+cycler @ file:///tmp/build/80754af9/cycler_1637851556182/work
+Cython @ file:///opt/conda/conda-bld/cython_1663692770955/work
+cytoolz==0.11.0
+daal4py==2021.6.0
+dask @ file:///tmp/abs_994957d9-ec12-411f-b953-c010f9d489d10hj3gz4k/croots/recipe/dask-core_1658513209934/work
+datashader @ file:///tmp/abs_aa58dfo4_s/croots/recipe/datashader_1659349033064/work
+datashape==0.5.4
+debugpy @ file:///tmp/build/80754af9/debugpy_1637091799509/work
+decorator @ file:///opt/conda/conda-bld/decorator_1643638310831/work
+defusedxml @ file:///tmp/build/80754af9/defusedxml_1615228127516/work
+diff-match-patch @ file:///Users/ktietz/demo/mc3/conda-bld/diff-match-patch_1630511840874/work
+dill==0.3.7
+distributed @ file:///tmp/abs_593da390-bd12-4acc-ba49-4c9993cbe8abgqg_w3rb/croots/recipe/distributed_1658520746481/work
+docutils @ file:///opt/conda/conda-bld/docutils_1657175430858/work
+entrypoints @ file:///tmp/build/80754af9/entrypoints_1649926439650/work
+et-xmlfile==1.1.0
+exceptiongroup==1.1.3
+-e git+ssh://[email protected]/X1AOX1A/Demos.git@bddbd1d412227d9a27b0e3dc7c7fa9e5e17a614a#egg=fairseq&subdirectory=fairseq
+fastapi==0.103.1
+fastjsonschema @ file:///opt/conda/conda-bld/python-fastjsonschema_1661371079312/work
+ffmpy==0.3.1
+filelock @ file:///opt/conda/conda-bld/filelock_1647002191454/work
+flake8 @ file:///opt/conda/conda-bld/flake8_1648129545443/work
+Flask @ file:///home/ktietz/src/ci/flask_1611932660458/work
+flatbuffers==23.5.9
+fonttools==4.25.0
+fsspec @ file:///opt/conda/conda-bld/fsspec_1659972197723/work
+future @ file:///tmp/build/80754af9/future_1607571303524/work
+gast==0.4.0
+gensim @ file:///tmp/build/80754af9/gensim_1646806807927/work
+gitdb==4.0.10
+GitPython==3.1.32
+glob2 @ file:///home/linux1/recipes/ci/glob2_1610991677669/work
+gmpy2 @ file:///tmp/build/80754af9/gmpy2_1645438755360/work
+google-auth==2.18.0
+google-auth-oauthlib==1.0.0
+google-pasta==0.2.0
+gradio==3.43.2
+gradio_client==0.5.0
+greenlet @ file:///tmp/build/80754af9/greenlet_1628888132713/work
+grpcio==1.54.0
+h11==0.14.0
+h5py @ file:///tmp/abs_4aewd3wzey/croots/recipe/h5py_1659091371897/work
+HeapDict @ file:///Users/ktietz/demo/mc3/conda-bld/heapdict_1630598515714/work
+holoviews @ file:///tmp/abs_eecc808c-455e-4be4-9911-ecf8341b3a34jfwskiqe/croots/recipe/holoviews_1658171506757/work
+httpcore==0.17.3
+httpx==0.24.1
+huggingface-hub==0.16.4
+hvplot @ file:///tmp/abs_6fcys5jcv1/croots/recipe/hvplot_1659026496554/work
+hydra-core==1.0.7
+hyperlink @ file:///tmp/build/80754af9/hyperlink_1610130746837/work
+idna @ file:///tmp/build/80754af9/idna_1637925883363/work
+imagecodecs @ file:///opt/conda/conda-bld/imagecodecs_1664561985385/work
+imageio @ file:///tmp/abs_cd920173-f360-47c5-97b0-bf4d1076d5d4dvic0oys/croots/recipe/imageio_1658785036907/work
+imagesize @ file:///opt/conda/conda-bld/imagesize_1657179498843/work
+importlib-metadata @ file:///tmp/build/80754af9/importlib-metadata_1648544546694/work
+importlib-resources==6.0.1
+incremental @ file:///tmp/build/80754af9/incremental_1636629750599/work
+inflection==0.5.1
+iniconfig @ file:///home/linux1/recipes/ci/iniconfig_1610983019677/work
+intake @ file:///opt/conda/conda-bld/intake_1647436631684/work
+intervaltree @ file:///Users/ktietz/demo/mc3/conda-bld/intervaltree_1630511889664/work
+ipykernel @ file:///opt/conda/conda-bld/ipykernel_1662361798230/work
+ipython @ file:///tmp/abs_94gruux8u8/croots/recipe/ipython_1659529858706/work
+ipython-genutils @ file:///tmp/build/80754af9/ipython_genutils_1606773439826/work
+ipywidgets @ file:///tmp/build/80754af9/ipywidgets_1634143127070/work
+isort @ file:///tmp/build/80754af9/isort_1628603791788/work
+itemadapter @ file:///tmp/build/80754af9/itemadapter_1626442940632/work
+itemloaders @ file:///opt/conda/conda-bld/itemloaders_1646805235997/work
+itsdangerous @ file:///tmp/build/80754af9/itsdangerous_1621432558163/work
+jax==0.4.9
+jdcal @ file:///Users/ktietz/demo/mc3/conda-bld/jdcal_1630584345063/work
+jedi @ file:///tmp/build/80754af9/jedi_1644297102865/work
+jeepney @ file:///tmp/build/80754af9/jeepney_1627537048313/work
+jellyfish @ file:///tmp/build/80754af9/jellyfish_1647944426575/work
+Jinja2 @ file:///tmp/build/80754af9/jinja2_1612213139570/work
+jinja2-time @ file:///opt/conda/conda-bld/jinja2-time_1649251842261/work
+jmespath @ file:///Users/ktietz/demo/mc3/conda-bld/jmespath_1630583964805/work
+joblib @ file:///tmp/build/80754af9/joblib_1635411271373/work
+json5 @ file:///tmp/build/80754af9/json5_1624432770122/work
+jsonschema @ file:///opt/conda/conda-bld/jsonschema_1663375472438/work
+jupyter @ file:///tmp/abs_33h4eoipez/croots/recipe/jupyter_1659349046347/work
+jupyter-console @ file:///opt/conda/conda-bld/jupyter_console_1647002188872/work
+jupyter-contrib-core==0.4.2
+jupyter-contrib-nbextensions==0.7.0
+jupyter-highlight-selected-word==0.2.0
+jupyter-nbextensions-configurator==0.6.1
+jupyter-server @ file:///tmp/abs_b88b31b8-83b9-476d-a46d-e563c421f38fvsnyi1ur/croots/recipe/jupyter_server_1658754481507/work
+jupyter_client @ file:///opt/conda/conda-bld/jupyter_client_1661848916004/work
+jupyter_core @ file:///opt/conda/conda-bld/jupyter_core_1664917302524/work
+jupyterlab @ file:///tmp/abs_12f3h01vmy/croots/recipe/jupyterlab_1658907535764/work
+jupyterlab-pygments @ file:///tmp/build/80754af9/jupyterlab_pygments_1601490720602/work
+jupyterlab-server @ file:///opt/conda/conda-bld/jupyterlab_server_1644500396812/work
+jupyterlab-widgets @ file:///tmp/build/80754af9/jupyterlab_widgets_1609884341231/work
+keras==2.12.0
+keyring @ file:///tmp/build/80754af9/keyring_1638531355686/work
+kiwisolver @ file:///opt/conda/conda-bld/kiwisolver_1653292039266/work
+lazy-object-proxy @ file:///tmp/build/80754af9/lazy-object-proxy_1616529027849/work
+libarchive-c @ file:///tmp/build/80754af9/python-libarchive-c_1617780486945/work
+libclang==16.0.0
+lit==16.0.3
+llvmlite==0.38.0
+locket @ file:///opt/conda/conda-bld/locket_1652903118915/work
+loguru==0.7.0
+lxml @ file:///opt/conda/conda-bld/lxml_1657545139709/work
+lz4 @ file:///tmp/build/80754af9/lz4_1619516502891/work
+Markdown @ file:///tmp/build/80754af9/markdown_1614363852612/work
+markdown-it-py==3.0.0
+MarkupSafe @ file:///tmp/build/80754af9/markupsafe_1621523467000/work
+matplotlib @ file:///opt/conda/conda-bld/matplotlib-suite_1660167928326/work
+matplotlib-inline @ file:///opt/conda/conda-bld/matplotlib-inline_1662014470464/work
+mccabe @ file:///opt/conda/conda-bld/mccabe_1644221741721/work
+mdurl==0.1.2
+-e git+ssh://[email protected]/X1AOX1A/Constrained_Image_Caption.git@36bce6f4b517b8f5ea26ec00a171523ffd0fa874#egg=minigpt4
+mistune @ file:///tmp/build/80754af9/mistune_1607364877025/work
+mkl-fft==1.3.1
+mkl-random @ file:///tmp/build/80754af9/mkl_random_1626186066731/work
+mkl-service==2.4.0
+ml-dtypes==0.1.0
+mock @ file:///tmp/build/80754af9/mock_1607622725907/work
+mpmath==1.2.1
+msgpack @ file:///opt/conda/conda-bld/msgpack-python_1652362659880/work
+multipledispatch @ file:///tmp/build/80754af9/multipledispatch_1607574243360/work
+multiprocess==0.70.15
+munkres==1.1.4
+mypy-extensions==0.4.3
+navigator-updater==0.3.0
+nbclassic @ file:///opt/conda/conda-bld/nbclassic_1644943264176/work
+nbclient @ file:///tmp/build/80754af9/nbclient_1650290509967/work
+nbconvert @ file:///opt/conda/conda-bld/nbconvert_1649751911790/work
+nbformat @ file:///opt/conda/conda-bld/nbformat_1663744952973/work
+nest-asyncio @ file:///tmp/build/80754af9/nest-asyncio_1649847906199/work
+networkx @ file:///opt/conda/conda-bld/networkx_1657784097507/work
+nltk @ file:///opt/conda/conda-bld/nltk_1645628263994/work
+nose @ file:///opt/conda/conda-bld/nose_1642704612149/work
+notebook @ file:///tmp/abs_abf6xa6h6f/croots/recipe/notebook_1659083654985/work
+numba @ file:///opt/conda/conda-bld/numba_1648040517072/work
+numexpr @ file:///opt/conda/conda-bld/numexpr_1656940300424/work
+numpy==1.23.5
+numpydoc @ file:///opt/conda/conda-bld/numpydoc_1657529872251/work
+nvidia-cublas-cu11==11.10.3.66
+nvidia-cuda-cupti-cu11==11.7.101
+nvidia-cuda-nvrtc-cu11==11.7.99
+nvidia-cuda-runtime-cu11==11.7.99
+nvidia-cudnn-cu11==8.5.0.96
+nvidia-cufft-cu11==10.9.0.58
+nvidia-curand-cu11==10.2.10.91
+nvidia-cusolver-cu11==11.4.0.1
+nvidia-cusparse-cu11==11.7.4.91
+nvidia-nccl-cu11==2.14.3
+nvidia-nvtx-cu11==11.7.91
+oauthlib==3.2.2
+olefile @ file:///Users/ktietz/demo/mc3/conda-bld/olefile_1629805411829/work
+omegaconf==2.0.6
+opencv-python==4.8.0.76
+openpyxl==3.0.10
+opt-einsum==3.3.0
+orjson==3.9.7
+packaging @ file:///tmp/build/80754af9/packaging_1637314298585/work
+pandas==1.4.4
+pandocfilters @ file:///opt/conda/conda-bld/pandocfilters_1643405455980/work
+panel @ file:///tmp/abs_bb3d3b2f-b3ea-41c0-a72e-8f54852d5cdfs70inytz/croots/recipe/panel_1658133826470/work
+param @ file:///tmp/build/80754af9/param_1636647414893/work
+parsel @ file:///tmp/build/80754af9/parsel_1646722533460/work
+parso @ file:///opt/conda/conda-bld/parso_1641458642106/work
+partd @ file:///opt/conda/conda-bld/partd_1647245470509/work
+pathlib @ file:///Users/ktietz/demo/mc3/conda-bld/pathlib_1629713961906/work
+pathos==0.3.1
+pathspec @ file:///tmp/abs_1foqurpsov/croots/recipe/pathspec_1659627126545/work
+patsy==0.5.2
+pep8==1.7.1
+pexpect @ file:///tmp/build/80754af9/pexpect_1605563209008/work
+pickleshare @ file:///tmp/build/80754af9/pickleshare_1606932040724/work
+Pillow==9.2.0
+pkginfo @ file:///tmp/build/80754af9/pkginfo_1643162084911/work
+platformdirs @ file:///opt/conda/conda-bld/platformdirs_1662711380096/work
+plotly @ file:///tmp/abs_7afcdfad-dbbb-49d2-adea-186abf525c45jbnd8p95/croots/recipe/plotly_1658160053621/work
+pluggy @ file:///tmp/build/80754af9/pluggy_1648024445381/work
+ply==3.11
+portalocker==2.7.0
+pox==0.3.3
+poyo @ file:///tmp/build/80754af9/poyo_1617751526755/work
+ppft==1.7.6.7
+prometheus-client @ file:///tmp/abs_d3zeliano1/croots/recipe/prometheus_client_1659455100375/work
+prompt-toolkit @ file:///tmp/build/80754af9/prompt-toolkit_1633440160888/work
+Protego @ file:///tmp/build/80754af9/protego_1598657180827/work
+protobuf==4.23.0
+psutil @ file:///opt/conda/conda-bld/psutil_1656431268089/work
+ptyprocess @ file:///tmp/build/80754af9/ptyprocess_1609355006118/work/dist/ptyprocess-0.7.0-py2.py3-none-any.whl
+py @ file:///opt/conda/conda-bld/py_1644396412707/work
+pyarrow==12.0.1
+pyasn1 @ file:///Users/ktietz/demo/mc3/conda-bld/pyasn1_1629708007385/work
+pyasn1-modules==0.2.8
+pycocotools==2.0.6
+pycodestyle @ file:///tmp/build/80754af9/pycodestyle_1636635402688/work
+pycosat==0.6.3
+pycparser @ file:///tmp/build/80754af9/pycparser_1636541352034/work
+pyct @ file:///tmp/abs_68a517ee-55fb-480e-82ab-1a8adb440a58x7qfc024/croots/recipe/pyct_1658500310800/work
+pycurl==7.45.1
+pydantic==2.3.0
+pydantic_core==2.6.3
+pydeck==0.8.0
+PyDispatcher==2.0.5
+pydocstyle @ file:///tmp/build/80754af9/pydocstyle_1621600989141/work
+pydub==0.25.1
+pyerfa @ file:///tmp/build/80754af9/pyerfa_1621556109336/work
+pyflakes @ file:///tmp/build/80754af9/pyflakes_1636644436481/work
+Pygments==2.16.1
+PyHamcrest @ file:///tmp/build/80754af9/pyhamcrest_1615748656804/work
+PyJWT @ file:///opt/conda/conda-bld/pyjwt_1657544592787/work
+pylint @ file:///tmp/abs_6fxmc66kyk/croots/recipe/pylint_1659110350161/work
+pyls-spyder==0.4.0
+Pympler==1.0.1
+pyodbc @ file:///tmp/abs_d365zrcsdp/croots/recipe/pyodbc_1659513794382/work
+pyOpenSSL @ file:///opt/conda/conda-bld/pyopenssl_1643788558760/work
+pyparsing @ file:///opt/conda/conda-bld/pyparsing_1661452539315/work
+PyQt5-sip==12.11.0
+pyrsistent @ file:///tmp/build/80754af9/pyrsistent_1636110951836/work
+PySocks @ file:///tmp/build/80754af9/pysocks_1605305812635/work
+pytest==7.1.2
+python-dateutil @ file:///tmp/build/80754af9/python-dateutil_1626374649649/work
+python-lsp-black @ file:///opt/conda/conda-bld/python-lsp-black_1661852031497/work
+python-lsp-jsonrpc==1.0.0
+python-lsp-server @ file:///opt/conda/conda-bld/python-lsp-server_1661813814476/work
+python-multipart==0.0.6
+python-slugify @ file:///tmp/build/80754af9/python-slugify_1620405669636/work
+python-snappy @ file:///tmp/build/80754af9/python-snappy_1610133040135/work
+pytz @ file:///opt/conda/conda-bld/pytz_1654762638606/work
+pytz-deprecation-shim==0.1.0.post0
+pyviz-comms @ file:///tmp/build/80754af9/pyviz_comms_1623747165329/work
+PyWavelets @ file:///tmp/build/80754af9/pywavelets_1648710015787/work
+pyxdg @ file:///tmp/build/80754af9/pyxdg_1603822279816/work
+PyYAML==6.0
+pyzmq @ file:///opt/conda/conda-bld/pyzmq_1657724186960/work
+QDarkStyle @ file:///tmp/build/80754af9/qdarkstyle_1617386714626/work
+qstylizer @ file:///tmp/build/80754af9/qstylizer_1617713584600/work/dist/qstylizer-0.1.10-py2.py3-none-any.whl
+QtAwesome @ file:///tmp/build/80754af9/qtawesome_1637160816833/work
+qtconsole @ file:///opt/conda/conda-bld/qtconsole_1662018252641/work
+QtPy @ file:///opt/conda/conda-bld/qtpy_1662014892439/work
+queuelib==1.5.0
+regex @ file:///tmp/abs_41f5bce5-0a2e-45aa-b231-1fd2fbd57753gfpe6sjm/croots/recipe/regex_1658257178822/work
+requests @ file:///opt/conda/conda-bld/requests_1657734628632/work
+requests-file @ file:///Users/ktietz/demo/mc3/conda-bld/requests-file_1629455781986/work
+requests-oauthlib==1.3.1
+rich==13.5.2
+rope @ file:///opt/conda/conda-bld/rope_1643788605236/work
+rsa==4.9
+Rtree @ file:///tmp/build/80754af9/rtree_1618420843093/work
+ruamel-yaml-conda @ file:///tmp/build/80754af9/ruamel_yaml_1616016711199/work
+ruamel.yaml @ file:///croot/ruamel.yaml_1666304550667/work
+ruamel.yaml.clib @ file:///croot/ruamel.yaml.clib_1666302247304/work
+s3transfer @ file:///opt/conda/conda-bld/s3transfer_1654524197066/work
+sacrebleu==2.3.1
+sacremoses==0.0.53
+scikit-image @ file:///tmp/build/80754af9/scikit-image_1648214171611/work
+scikit-learn @ file:///tmp/build/80754af9/scikit-learn_1642617106979/work
+scikit-learn-intelex==2021.20221004.171807
+scipy==1.9.1
+Scrapy @ file:///tmp/abs_e3bmwi01y8/croots/recipe/scrapy_1659598696235/work
+seaborn @ file:///tmp/build/80754af9/seaborn_1629307859561/work
+SecretStorage @ file:///tmp/build/80754af9/secretstorage_1614022780358/work
+semantic-version==2.10.0
+Send2Trash @ file:///tmp/build/80754af9/send2trash_1632406701022/work
+service-identity @ file:///Users/ktietz/demo/mc3/conda-bld/service_identity_1629460757137/work
+sip @ file:///tmp/abs_44cd77b_pu/croots/recipe/sip_1659012365470/work
+six @ file:///tmp/build/80754af9/six_1644875935023/work
+smart-open @ file:///opt/conda/conda-bld/smart_open_1651563547610/work
+smmap==5.0.0
+sniffio @ file:///tmp/build/80754af9/sniffio_1614030464178/work
+snowballstemmer @ file:///tmp/build/80754af9/snowballstemmer_1637937080595/work
+sortedcollections @ file:///tmp/build/80754af9/sortedcollections_1611172717284/work
+sortedcontainers @ file:///tmp/build/80754af9/sortedcontainers_1623949099177/work
+soupsieve @ file:///tmp/build/80754af9/soupsieve_1636706018808/work
+speaksee==0.0.1
+Sphinx @ file:///opt/conda/conda-bld/sphinx_1657784123546/work
+sphinxcontrib-applehelp @ file:///home/ktietz/src/ci/sphinxcontrib-applehelp_1611920841464/work
+sphinxcontrib-devhelp @ file:///home/ktietz/src/ci/sphinxcontrib-devhelp_1611920923094/work
+sphinxcontrib-htmlhelp @ file:///tmp/build/80754af9/sphinxcontrib-htmlhelp_1623945626792/work
+sphinxcontrib-jsmath @ file:///home/ktietz/src/ci/sphinxcontrib-jsmath_1611920942228/work
+sphinxcontrib-qthelp @ file:///home/ktietz/src/ci/sphinxcontrib-qthelp_1611921055322/work
+sphinxcontrib-serializinghtml @ file:///tmp/build/80754af9/sphinxcontrib-serializinghtml_1624451540180/work
+spyder @ file:///opt/conda/conda-bld/spyder_1663056818299/work
+spyder-kernels @ file:///opt/conda/conda-bld/spyder-kernels_1662457880976/work
+SQLAlchemy @ file:///tmp/abs_18b3238f-9c23-4182-a392-63af30a93c1er8j_yw60/croots/recipe/sqlalchemy_1657867856580/work
+starlette==0.27.0
+statsmodels @ file:///tmp/build/80754af9/statsmodels_1648015433305/work
+streamlit==1.25.0
+streamlit-drawable-canvas==0.9.3
+sympy @ file:///tmp/build/80754af9/sympy_1647853653589/work
+tables @ file:///tmp/build/80754af9/pytables_1607975397488/work
+tabulate @ file:///opt/conda/conda-bld/tabulate_1657784105888/work
+TBB==0.2
+tblib @ file:///Users/ktietz/demo/mc3/conda-bld/tblib_1629402031467/work
+tenacity==8.2.2
+tensorboard==2.12.3
+tensorboard-data-server==0.7.0
+tensorboardX==2.6.2
+tensorflow==2.12.0
+tensorflow-estimator==2.12.0
+tensorflow-io-gcs-filesystem==0.32.0
+termcolor==2.3.0
+terminado @ file:///tmp/build/80754af9/terminado_1644322582718/work
+testpath @ file:///opt/conda/conda-bld/testpath_1655908557405/work
+text-unidecode @ file:///Users/ktietz/demo/mc3/conda-bld/text-unidecode_1629401354553/work
+textdistance @ file:///tmp/build/80754af9/textdistance_1612461398012/work
+threadpoolctl @ file:///Users/ktietz/demo/mc3/conda-bld/threadpoolctl_1629802263681/work
+three-merge @ file:///tmp/build/80754af9/three-merge_1607553261110/work
+tifffile @ file:///tmp/build/80754af9/tifffile_1627275862826/work
+tinycss @ file:///tmp/build/80754af9/tinycss_1617713798712/work
+tldextract @ file:///opt/conda/conda-bld/tldextract_1646638314385/work
+toml @ file:///tmp/build/80754af9/toml_1616166611790/work
+tomli @ file:///opt/conda/conda-bld/tomli_1657175507142/work
+tomlkit @ file:///tmp/abs_56_0lnnq5x/croots/recipe/tomlkit_1658946880479/work
+toolz @ file:///tmp/build/80754af9/toolz_1636545406491/work
+torch==2.0.1
+torchaudio==2.0.1
+torchvision==0.15.2
+tornado @ file:///tmp/build/80754af9/tornado_1606942317143/work
+tqdm @ file:///opt/conda/conda-bld/tqdm_1664392687731/work
+traitlets @ file:///tmp/build/80754af9/traitlets_1636710298902/work
+triton==2.0.0
+Twisted @ file:///tmp/abs_82802zpkox/croots/recipe/twisted_1659592759417/work
+typing_extensions==4.7.1
+tzdata==2023.3
+tzlocal==4.3.1
+ujson @ file:///opt/conda/conda-bld/ujson_1657544923770/work
+Unidecode @ file:///tmp/build/80754af9/unidecode_1614712377438/work
+urllib3 @ file:///tmp/abs_5dhwnz6atv/croots/recipe/urllib3_1659110457909/work
+uvicorn==0.23.2
+validators==0.20.0
+w3lib @ file:///Users/ktietz/demo/mc3/conda-bld/w3lib_1629359764703/work
+watchdog @ file:///tmp/build/80754af9/watchdog_1638367282716/work
+wcwidth @ file:///Users/ktietz/demo/mc3/conda-bld/wcwidth_1629357192024/work
+webencodings==0.5.1
+websocket-client @ file:///tmp/build/80754af9/websocket-client_1614803975924/work
+websockets==11.0.3
+Werkzeug @ file:///opt/conda/conda-bld/werkzeug_1645628268370/work
+whatthepatch @ file:///opt/conda/conda-bld/whatthepatch_1661795988879/work
+widgetsnbextension @ file:///tmp/build/80754af9/widgetsnbextension_1644992802045/work
+wrapt @ file:///tmp/abs_c335821b-6e43-4504-9816-b1a52d3d3e1eel6uae8l/croots/recipe/wrapt_1657814400492/work
+wurlitzer @ file:///tmp/build/80754af9/wurlitzer_1638368168359/work
+xarray @ file:///opt/conda/conda-bld/xarray_1639166117697/work
+xlrd @ file:///tmp/build/80754af9/xlrd_1608072521494/work
+XlsxWriter @ file:///opt/conda/conda-bld/xlsxwriter_1649073856329/work
+yapf @ file:///tmp/build/80754af9/yapf_1615749224965/work
+zict==2.1.0
+zipp @ file:///opt/conda/conda-bld/zipp_1652341764480/work
+zope.interface @ file:///tmp/build/80754af9/zope.interface_1625036153595/work

utils.py

@@ -0,0 +1,17 @@
+import jieba
+from functools import partial
+from transformers import BertTokenizer
+
+class T5PegasusTokenizer(BertTokenizer):
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        self.pre_tokenizer = partial(jieba.cut, HMM=False)
+
+    def _tokenize(self, text, *arg, **kwargs):
+        split_tokens = []
+        for text in self.pre_tokenizer(text):
+            if text in self.vocab:
+                split_tokens.append(text)
+            else:
+                split_tokens.extend(super()._tokenize(text))
+        return split_tokens