গ্রাউন্ডেড_স্ক্যান

  • বর্ণনা :

গ্রাউন্ডেড স্ক্যান (gSCAN) হল একটি সিন্থেটিক ডেটাসেট যা অবস্থানগত ভাষা বোঝার ক্ষেত্রে রচনাগত সাধারণীকরণের মূল্যায়ন করার জন্য। gSCAN প্রাকৃতিক ভাষার নির্দেশাবলীকে অ্যাকশন সিকোয়েন্সের সাথে যুক্ত করে এবং এজেন্টকে একটি গ্রিড-ভিত্তিক ভিজ্যুয়াল নেভিগেশন পরিবেশের প্রেক্ষাপটে নির্দেশাবলী ব্যাখ্যা করতে চায়।

আরও তথ্য এখানে পাওয়া যাবে:

FeaturesDict({
    'command': Sequence(Text(shape=(), dtype=string)),
    'manner': Text(shape=(), dtype=string),
    'meaning': Sequence(Text(shape=(), dtype=string)),
    'referred_target': Text(shape=(), dtype=string),
    'situation': FeaturesDict({
        'agent_direction': int32,
        'agent_position': FeaturesDict({
            'column': int32,
            'row': int32,
        }),
        'direction_to_target': Text(shape=(), dtype=string),
        'distance_to_target': int32,
        'grid_size': int32,
        'placed_objects': Sequence({
            'object': FeaturesDict({
                'color': Text(shape=(), dtype=string),
                'shape': Text(shape=(), dtype=string),
                'size': int32,
            }),
            'position': FeaturesDict({
                'column': int32,
                'row': int32,
            }),
            'vector': Text(shape=(), dtype=string),
        }),
        'target_object': FeaturesDict({
            'object': FeaturesDict({
                'color': Text(shape=(), dtype=string),
                'shape': Text(shape=(), dtype=string),
                'size': int32,
            }),
            'position': FeaturesDict({
                'column': int32,
                'row': int32,
            }),
            'vector': Text(shape=(), dtype=string),
        }),
    }),
    'target_commands': Sequence(Text(shape=(), dtype=string)),
    'verb_in_command': Text(shape=(), dtype=string),
})
  • বৈশিষ্ট্য ডকুমেন্টেশন :
বৈশিষ্ট্য ক্লাস আকৃতি ডিটাইপ বর্ণনা
ফিচারসডিক্ট
আদেশ ক্রম (পাঠ্য) (কোনটিই নয়,) স্ট্রিং
পদ্ধতি পাঠ্য স্ট্রিং
অর্থ ক্রম (পাঠ্য) (কোনটিই নয়,) স্ট্রিং
উল্লেখিত_লক্ষ্য পাঠ্য স্ট্রিং
অবস্থা ফিচারসডিক্ট
পরিস্থিতি/এজেন্ট_নির্দেশ টেনসর int32
পরিস্থিতি/এজেন্ট_অবস্থান ফিচারসডিক্ট
পরিস্থিতি/এজেন্ট_অবস্থান/কলাম টেনসর int32
পরিস্থিতি/এজেন্ট_অবস্থান/সারি টেনসর int32
পরিস্থিতি/নির্দেশ_প্রতি_লক্ষ্য পাঠ্য স্ট্রিং
পরিস্থিতি/দূরত্ব_থেকে_লক্ষ্য টেনসর int32
পরিস্থিতি/গ্রিড_সাইজ টেনসর int32
পরিস্থিতি/স্থাপিত_বস্তু ক্রম
পরিস্থিতি/স্থাপিত_বস্তু/বস্তু ফিচারসডিক্ট
পরিস্থিতি/স্থাপিত_বস্তু/বস্তু/রঙ পাঠ্য স্ট্রিং
পরিস্থিতি/স্থাপিত_বস্তু/বস্তু/আকৃতি পাঠ্য স্ট্রিং
পরিস্থিতি/স্থাপিত_বস্তু/বস্তু/আকার টেনসর int32
পরিস্থিতি/স্থাপিত_বস্তু/অবস্থান ফিচারসডিক্ট
পরিস্থিতি/স্থাপিত_বস্তু/অবস্থান/কলাম টেনসর int32
পরিস্থিতি/স্থাপিত_বস্তু/অবস্থান/সারি টেনসর int32
পরিস্থিতি/স্থাপিত_বস্তু/ভেক্টর পাঠ্য স্ট্রিং
পরিস্থিতি/লক্ষ্য_বস্তু ফিচারসডিক্ট
পরিস্থিতি/টার্গেট_অবজেক্ট/অবজেক্ট ফিচারসডিক্ট
পরিস্থিতি/লক্ষ্য_বস্তু/বস্তু/রঙ পাঠ্য স্ট্রিং
পরিস্থিতি/লক্ষ্য_বস্তু/বস্তু/আকৃতি পাঠ্য স্ট্রিং
পরিস্থিতি/লক্ষ্য_বস্তু/বস্তু/আকার টেনসর int32
পরিস্থিতি/লক্ষ্য_বস্তু/অবস্থান ফিচারসডিক্ট
পরিস্থিতি/টার্গেট_অবজেক্ট/অবস্থান/কলাম টেনসর int32
পরিস্থিতি/লক্ষ্য_বস্তু/অবস্থান/সারি টেনসর int32
পরিস্থিতি/লক্ষ্য_বস্তু/ভেক্টর পাঠ্য স্ট্রিং
টার্গেট_কমান্ড ক্রম (পাঠ্য) (কোনটিই নয়,) স্ট্রিং
verb_in_command পাঠ্য স্ট্রিং
  • তত্ত্বাবধান করা কী (দেখুন as_supervised doc ): None

  • চিত্র ( tfds.show_examples ): সমর্থিত নয়।

  • উদ্ধৃতি :

@inproceedings{NEURIPS2020_e5a90182,
 author = {Ruis, Laura and Andreas, Jacob and Baroni, Marco and Bouchacourt, Diane and Lake, Brenden M},
 booktitle = {Advances in Neural Information Processing Systems},
 editor = {H. Larochelle and M. Ranzato and R. Hadsell and M. F. Balcan and H. Lin},
 pages = {19861--19872},
 publisher = {Curran Associates, Inc.},
 title = {A Benchmark for Systematic Generalization in Grounded Language Understanding},
 url = {https://proceedings.neurips.cc/paper/2020/file/e5a90182cc81e12ab5e72d66e0b46fe3-Paper.pdf},
 volume = {33},
 year = {2020}
}

@inproceedings{qiu-etal-2021-systematic,
    title = "Systematic Generalization on g{SCAN}: {W}hat is Nearly Solved and What is Next?",
    author = "Qiu, Linlu  and
      Hu, Hexiang  and
      Zhang, Bowen  and
      Shaw, Peter  and
      Sha, Fei",
    booktitle = "Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing",
    month = nov,
    year = "2021",
    address = "Online and Punta Cana, Dominican Republic",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2021.emnlp-main.166",
    doi = "10.18653/v1/2021.emnlp-main.166",
    pages = "2180--2188",
}

grounded_scan/compositional_splits (ডিফল্ট কনফিগারেশন)

  • কনফিগারেশনের বিবরণ : রচনামূলক সাধারণীকরণের উদাহরণ।

  • ডাউনলোড সাইজ : 82.10 MiB

  • ডেটাসেটের আকার : 998.11 MiB

  • বিভাজন :

বিভক্ত উদাহরণ
'adverb_1' 112,880
'adverb_2' 38,582
'contextual' 11,460
'dev' 3,716
'situational_1' ৮৮,৬৪২
'situational_2' 16,808
'test' 19,282
'train' 367,933
'visual' 37,436
'visual_easier' 18,718

গ্রাউন্ডেড_স্ক্যান/টার্গেট_লেংথ_বিভাজন

  • কনফিগার বিবরণ : বড় লক্ষ্য দৈর্ঘ্যের সাধারণীকরণের উদাহরণ।

  • ডাউনলোড সাইজ : 53.41 MiB

  • ডেটাসেটের আকার : 546.73 MiB

  • বিভাজন :

বিভক্ত উদাহরণ
'dev' 1,821
'target_lengths' 198,588
'test' 37,784
'train' 180,301

স্থল_স্ক্যান/স্থানিক_সম্পর্ক_বিভাজন

  • কনফিগার বর্ণনা : স্থানিক সম্পর্ক যুক্তির উদাহরণ।

  • ডাউনলোড সাইজ : 89.59 MiB

  • ডেটাসেটের আকার : 675.09 MiB

  • বিভাজন :

বিভক্ত উদাহরণ
'dev' 2,617
'referent' 30,492
'relation' 6,285
'relative_position_1' 41,576
'relative_position_2' 41,529
'test' 28,526
'train' 259,088
'visual' ৬২,২৫০