feat: integrate TextToAMR

app.py

@@ -1,8 +1,28 @@
+from git import Repo
 import gradio as gr
+from huggingface_hub import snapshot_download
 import penman
+import sys
+
+Repo.clone_from("https://github.com/AbdiHaryadi/amr-tst-indo.git", "amr-tst-indo")
+sys.path.append("./amr-tst-indo")
+
+from text_to_amr import TextToAMR
+
+# Gold v2
+amr_parsing_model_name = "mbart-en-id-smaller-indo-amr-parsing-translated-nafkhan"
+snapshot_download(
+    repo_id=f"abdiharyadi/{amr_parsing_model_name}",
+    local_dir=f"./amr-tst-indo/AMRBART-id/models/{amr_parsing_model_name}",
+    ignore_patterns=[
+        "*log*",
+        "*checkpoint*",
+    ]
+)
+t2a = TextToAMR(model_name=amr_parsing_model_name)
 
 def run(text, source_style):
-    source_amr = penman.decode("(z0 / halo)")
+    source_amr, *_ = t2a([text])
     source_amr_display = penman.encode(source_amr)
     yield source_amr_display, "...", "...", "...", "..."
 

requirements.txt

@@ -1 +1,38 @@
+gitpython
+
+accelerate==0.32.1
+amrlib==0.8.0
+cached-property==1.5.2
+ConfigArgParse==1.7
+datasets==2.20.0
+editdistance==0.8.1
+evaluate==0.4.3
+fasttext==0.9.3
+h5py-cache==1.0
+networkx==3.3
+nltk==3.8.1
+pandas==2.2.2
 Penman==1.3.0
+pytorch-ignite==0.5.0.post2
+PyYAML==6.0.1
+regex==2024.5.15
+rouge_score==0.1.2
+sacrebleu==2.4.2
+sacremoses==0.1.1
+Sastrawi==1.0.1
+sentence-transformers==3.0.1
+sentencepiece==0.2.0
+smatch==1.0.4
+tensorboardX==2.6.2.2
+torch==2.4.0 --index-url https://download.pytorch.org/whl/cu121
+torchmetrics==1.4.0.post0
+tqdm==4.66.4
+transformers==4.44.0
+wandb==0.17.5
+
+# This library depends on what system you use, and doesn't impact the result.
+numpy
+torchvision --index-url https://download.pytorch.org/whl/cu121
+
+# This library version should be fixed until Indonesian stanza doesn't have any hash problem.
+stanza==1.4.2