added curriculum learning baseline

checkpoint/curriculum_learning_checkpoint.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c0c3e86df403cee2c091b1aab83070a0c2be2c71eb5bd71090a13f83501c36ee
+size 427253662

checkpoint/curriculum_learning_results/curriculum_learning_ewok_results.json

@@ -0,0 +1,450 @@
+{
+  "results": {
+    "ewok_filtered": {
+      "acc,none": 0.6136494390163773,
+      "acc_stderr,none": 0.004958271954057393,
+      "alias": "ewok_filtered"
+    },
+    "ewok_agent-properties_filtered": {
+      "acc,none": 0.7619909502262443,
+      "acc_stderr,none": 0.009060951046179946,
+      "alias": " - ewok_agent-properties_filtered"
+    },
+    "ewok_material-dynamics_filtered": {
+      "acc,none": 0.8272727272727273,
+      "acc_stderr,none": 0.013631437026268746,
+      "alias": " - ewok_material-dynamics_filtered"
+    },
+    "ewok_material-properties_filtered": {
+      "acc,none": 0.4176470588235294,
+      "acc_stderr,none": 0.03793625867642199,
+      "alias": " - ewok_material-properties_filtered"
+    },
+    "ewok_physical-dynamics_filtered": {
+      "acc,none": 0.5083333333333333,
+      "acc_stderr,none": 0.0458285584474836,
+      "alias": " - ewok_physical-dynamics_filtered"
+    },
+    "ewok_physical-interactions_filtered": {
+      "acc,none": 0.6025179856115108,
+      "acc_stderr,none": 0.020772906064232693,
+      "alias": " - ewok_physical-interactions_filtered"
+    },
+    "ewok_physical-relations_filtered": {
+      "acc,none": 0.8325183374083129,
+      "acc_stderr,none": 0.01306379953798592,
+      "alias": " - ewok_physical-relations_filtered"
+    },
+    "ewok_quantitative-properties_filtered": {
+      "acc,none": 0.4012738853503185,
+      "acc_stderr,none": 0.027705269581356752,
+      "alias": " - ewok_quantitative-properties_filtered"
+    },
+    "ewok_social-interactions_filtered": {
+      "acc,none": 0.5340136054421769,
+      "acc_stderr,none": 0.029142645214853977,
+      "alias": " - ewok_social-interactions_filtered"
+    },
+    "ewok_social-properties_filtered": {
+      "acc,none": 0.4573170731707317,
+      "acc_stderr,none": 0.027549131410256764,
+      "alias": " - ewok_social-properties_filtered"
+    },
+    "ewok_social-relations_filtered": {
+      "acc,none": 0.7990956072351422,
+      "acc_stderr,none": 0.010187061572694651,
+      "alias": " - ewok_social-relations_filtered"
+    },
+    "ewok_spatial-relations_filtered": {
+      "acc,none": 0.6081632653061224,
+      "acc_stderr,none": 0.022075382472922258,
+      "alias": " - ewok_spatial-relations_filtered"
+    }
+  },
+  "groups": {
+    "ewok_filtered": {
+      "acc,none": 0.6136494390163773,
+      "acc_stderr,none": 0.004958271954057393,
+      "alias": "ewok_filtered"
+    }
+  },
+  "group_subtasks": {
+    "ewok_filtered": [
+      "ewok_spatial-relations_filtered",
+      "ewok_social-relations_filtered",
+      "ewok_social-properties_filtered",
+      "ewok_social-interactions_filtered",
+      "ewok_quantitative-properties_filtered",
+      "ewok_physical-relations_filtered",
+      "ewok_physical-interactions_filtered",
+      "ewok_physical-dynamics_filtered",
+      "ewok_material-properties_filtered",
+      "ewok_material-dynamics_filtered",
+      "ewok_agent-properties_filtered"
+    ]
+  },
+  "configs": {
+    "ewok_agent-properties_filtered": {
+      "task": "ewok_agent-properties_filtered",
+      "group": "ewok_filtered",
+      "dataset_path": "json",
+      "dataset_kwargs": {
+        "data_files": "evaluation_data/ewok_filtered/agent-properties.jsonl"
+      },
+      "validation_split": "train",
+      "doc_to_text": "{{Context1}}",
+      "doc_to_target": 0,
+      "doc_to_choice": "{{[Target1, Target2]}}",
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "acc",
+          "weight_by_size": false
+        }
+      ],
+      "output_type": "multiple_choice",
+      "repeats": 1,
+      "should_decontaminate": true,
+      "doc_to_decontamination_query": "{{Context1}} {{Target1}}",
+      "metadata": {
+        "version": 1.0
+      }
+    },
+    "ewok_material-dynamics_filtered": {
+      "task": "ewok_material-dynamics_filtered",
+      "group": "ewok_filtered",
+      "dataset_path": "json",
+      "dataset_kwargs": {
+        "data_files": "evaluation_data/ewok_filtered/material-dynamics.jsonl"
+      },
+      "validation_split": "train",
+      "doc_to_text": "{{Context1}}",
+      "doc_to_target": 0,
+      "doc_to_choice": "{{[Target1, Target2]}}",
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "acc",
+          "weight_by_size": false
+        }
+      ],
+      "output_type": "multiple_choice",
+      "repeats": 1,
+      "should_decontaminate": true,
+      "doc_to_decontamination_query": "{{Context1}} {{Target1}}",
+      "metadata": {
+        "version": 1.0
+      }
+    },
+    "ewok_material-properties_filtered": {
+      "task": "ewok_material-properties_filtered",
+      "group": "ewok_filtered",
+      "dataset_path": "json",
+      "dataset_kwargs": {
+        "data_files": "evaluation_data/ewok_filtered/material-properties.jsonl"
+      },
+      "validation_split": "train",
+      "doc_to_text": "{{Context1}}",
+      "doc_to_target": 0,
+      "doc_to_choice": "{{[Target1, Target2]}}",
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "acc",
+          "weight_by_size": false
+        }
+      ],
+      "output_type": "multiple_choice",
+      "repeats": 1,
+      "should_decontaminate": true,
+      "doc_to_decontamination_query": "{{Context1}} {{Target1}}",
+      "metadata": {
+        "version": 1.0
+      }
+    },
+    "ewok_physical-dynamics_filtered": {
+      "task": "ewok_physical-dynamics_filtered",
+      "group": "ewok_filtered",
+      "dataset_path": "json",
+      "dataset_kwargs": {
+        "data_files": "evaluation_data/ewok_filtered/physical-dynamics.jsonl"
+      },
+      "validation_split": "train",
+      "doc_to_text": "{{Context1}}",
+      "doc_to_target": 0,
+      "doc_to_choice": "{{[Target1, Target2]}}",
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "acc",
+          "weight_by_size": false
+        }
+      ],
+      "output_type": "multiple_choice",
+      "repeats": 1,
+      "should_decontaminate": true,
+      "doc_to_decontamination_query": "{{Context1}} {{Target1}}",
+      "metadata": {
+        "version": 1.0
+      }
+    },
+    "ewok_physical-interactions_filtered": {
+      "task": "ewok_physical-interactions_filtered",
+      "group": "ewok_filtered",
+      "dataset_path": "json",
+      "dataset_kwargs": {
+        "data_files": "evaluation_data/ewok_filtered/physical-interactions.jsonl"
+      },
+      "validation_split": "train",
+      "doc_to_text": "{{Context1}}",
+      "doc_to_target": 0,
+      "doc_to_choice": "{{[Target1, Target2]}}",
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "acc",
+          "weight_by_size": false
+        }
+      ],
+      "output_type": "multiple_choice",
+      "repeats": 1,
+      "should_decontaminate": true,
+      "doc_to_decontamination_query": "{{Context1}} {{Target1}}",
+      "metadata": {
+        "version": 1.0
+      }
+    },
+    "ewok_physical-relations_filtered": {
+      "task": "ewok_physical-relations_filtered",
+      "group": "ewok_filtered",
+      "dataset_path": "json",
+      "dataset_kwargs": {
+        "data_files": "evaluation_data/ewok_filtered/physical-relations.jsonl"
+      },
+      "validation_split": "train",
+      "doc_to_text": "{{Context1}}",
+      "doc_to_target": 0,
+      "doc_to_choice": "{{[Target1, Target2]}}",
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "acc",
+          "weight_by_size": false
+        }
+      ],
+      "output_type": "multiple_choice",
+      "repeats": 1,
+      "should_decontaminate": true,
+      "doc_to_decontamination_query": "{{Context1}} {{Target1}}",
+      "metadata": {
+        "version": 1.0
+      }
+    },
+    "ewok_quantitative-properties_filtered": {
+      "task": "ewok_quantitative-properties_filtered",
+      "group": "ewok_filtered",
+      "dataset_path": "json",
+      "dataset_kwargs": {
+        "data_files": "evaluation_data/ewok_filtered/quantitative-properties.jsonl"
+      },
+      "validation_split": "train",
+      "doc_to_text": "{{Context1}}",
+      "doc_to_target": 0,
+      "doc_to_choice": "{{[Target1, Target2]}}",
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "acc",
+          "weight_by_size": false
+        }
+      ],
+      "output_type": "multiple_choice",
+      "repeats": 1,
+      "should_decontaminate": true,
+      "doc_to_decontamination_query": "{{Context1}} {{Target1}}",
+      "metadata": {
+        "version": 1.0
+      }
+    },
+    "ewok_social-interactions_filtered": {
+      "task": "ewok_social-interactions_filtered",
+      "group": "ewok_filtered",
+      "dataset_path": "json",
+      "dataset_kwargs": {
+        "data_files": "evaluation_data/ewok_filtered/social-interactions.jsonl"
+      },
+      "validation_split": "train",
+      "doc_to_text": "{{Context1}}",
+      "doc_to_target": 0,
+      "doc_to_choice": "{{[Target1, Target2]}}",
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "acc",
+          "weight_by_size": false
+        }
+      ],
+      "output_type": "multiple_choice",
+      "repeats": 1,
+      "should_decontaminate": true,
+      "doc_to_decontamination_query": "{{Context1}} {{Target1}}",
+      "metadata": {
+        "version": 1.0
+      }
+    },
+    "ewok_social-properties_filtered": {
+      "task": "ewok_social-properties_filtered",
+      "group": "ewok_filtered",
+      "dataset_path": "json",
+      "dataset_kwargs": {
+        "data_files": "evaluation_data/ewok_filtered/social-properties.jsonl"
+      },
+      "validation_split": "train",
+      "doc_to_text": "{{Context1}}",
+      "doc_to_target": 0,
+      "doc_to_choice": "{{[Target1, Target2]}}",
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "acc",
+          "weight_by_size": false
+        }
+      ],
+      "output_type": "multiple_choice",
+      "repeats": 1,
+      "should_decontaminate": true,
+      "doc_to_decontamination_query": "{{Context1}} {{Target1}}",
+      "metadata": {
+        "version": 1.0
+      }
+    },
+    "ewok_social-relations_filtered": {
+      "task": "ewok_social-relations_filtered",
+      "group": "ewok_filtered",
+      "dataset_path": "json",
+      "dataset_kwargs": {
+        "data_files": "evaluation_data/ewok_filtered/social-relations.jsonl"
+      },
+      "validation_split": "train",
+      "doc_to_text": "{{Context1}}",
+      "doc_to_target": 0,
+      "doc_to_choice": "{{[Target1, Target2]}}",
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "acc",
+          "weight_by_size": false
+        }
+      ],
+      "output_type": "multiple_choice",
+      "repeats": 1,
+      "should_decontaminate": true,
+      "doc_to_decontamination_query": "{{Context1}} {{Target1}}",
+      "metadata": {
+        "version": 1.0
+      }
+    },
+    "ewok_spatial-relations_filtered": {
+      "task": "ewok_spatial-relations_filtered",
+      "group": "ewok_filtered",
+      "dataset_path": "json",
+      "dataset_kwargs": {
+        "data_files": "evaluation_data/ewok_filtered/spatial-relations.jsonl"
+      },
+      "validation_split": "train",
+      "doc_to_text": "{{Context1}}",
+      "doc_to_target": 0,
+      "doc_to_choice": "{{[Target1, Target2]}}",
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "acc",
+          "weight_by_size": false
+        }
+      ],
+      "output_type": "multiple_choice",
+      "repeats": 1,
+      "should_decontaminate": true,
+      "doc_to_decontamination_query": "{{Context1}} {{Target1}}",
+      "metadata": {
+        "version": 1.0
+      }
+    }
+  },
+  "versions": {
+    "ewok_agent-properties_filtered": 1.0,
+    "ewok_material-dynamics_filtered": 1.0,
+    "ewok_material-properties_filtered": 1.0,
+    "ewok_physical-dynamics_filtered": 1.0,
+    "ewok_physical-interactions_filtered": 1.0,
+    "ewok_physical-relations_filtered": 1.0,
+    "ewok_quantitative-properties_filtered": 1.0,
+    "ewok_social-interactions_filtered": 1.0,
+    "ewok_social-properties_filtered": 1.0,
+    "ewok_social-relations_filtered": 1.0,
+    "ewok_spatial-relations_filtered": 1.0
+  },
+  "n-shot": {
+    "ewok_agent-properties_filtered": 0,
+    "ewok_filtered": 0,
+    "ewok_material-dynamics_filtered": 0,
+    "ewok_material-properties_filtered": 0,
+    "ewok_physical-dynamics_filtered": 0,
+    "ewok_physical-interactions_filtered": 0,
+    "ewok_physical-relations_filtered": 0,
+    "ewok_quantitative-properties_filtered": 0,
+    "ewok_social-interactions_filtered": 0,
+    "ewok_social-properties_filtered": 0,
+    "ewok_social-relations_filtered": 0,
+    "ewok_spatial-relations_filtered": 0
+  },
+  "config": {
+    "model": "hf-mlm",
+    "model_args": "pretrained=../models/ELC_ParserBERT_10M_256_ordered/,backend=mlm,trust_remote_code=True",
+    "batch_size": "128",
+    "batch_sizes": [],
+    "device": "cuda:0",
+    "use_cache": null,
+    "limit": null,
+    "bootstrap_iters": 100000,
+    "gen_kwargs": null
+  },
+  "git_hash": "0150eb5",
+  "date": 1726585565.004646,
+  "pretty_env_info": "PyTorch version: 2.4.1\nIs debug build: False\nCUDA used to build PyTorch: 12.1\nROCM used to build PyTorch: N/A\n\nOS: CentOS Linux release 7.9.2009 (Core) (x86_64)\nGCC version: (GCC) 4.8.5 20150623 (Red Hat 4.8.5-44)\nClang version: Could not collect\nCMake version: version 2.8.12.2\nLibc version: glibc-2.17\n\nPython version: 3.11.9 (main, Apr 19 2024, 16:48:06) [GCC 11.2.0] (64-bit runtime)\nPython platform: Linux-3.10.0-1160.71.1.el7.x86_64-x86_64-with-glibc2.17\nIs CUDA available: True\nCUDA runtime version: 12.1.105\nCUDA_MODULE_LOADING set to: LAZY\nGPU models and configuration: GPU 0: NVIDIA A100 80GB PCIe\nNvidia driver version: 545.23.08\ncuDNN version: Could not collect\nHIP runtime version: N/A\nMIOpen runtime version: N/A\nIs XNNPACK available: True\n\nCPU:\nArchitecture:          x86_64\nCPU op-mode(s):        32-bit, 64-bit\nByte Order:            Little Endian\nCPU(s):                128\nOn-line CPU(s) list:   0-127\nThread(s) per core:    2\nCore(s) per socket:    32\nSocket(s):             2\nNUMA node(s):          2\nVendor ID:             GenuineIntel\nCPU family:            6\nModel:                 106\nModel name:            Intel(R) Xeon(R) Platinum 8358 CPU @ 2.60GHz\nStepping:              6\nCPU MHz:               2600.000\nBogoMIPS:              5200.00\nVirtualization:        VT-x\nL1d cache:             48K\nL1i cache:             32K\nL2 cache:              1280K\nL3 cache:              49152K\nNUMA node0 CPU(s):     0,2,4,6,8,10,12,14,16,18,20,22,24,26,28,30,32,34,36,38,40,42,44,46,48,50,52,54,56,58,60,62,64,66,68,70,72,74,76,78,80,82,84,86,88,90,92,94,96,98,100,102,104,106,108,110,112,114,116,118,120,122,124,126\nNUMA node1 CPU(s):     1,3,5,7,9,11,13,15,17,19,21,23,25,27,29,31,33,35,37,39,41,43,45,47,49,51,53,55,57,59,61,63,65,67,69,71,73,75,77,79,81,83,85,87,89,91,93,95,97,99,101,103,105,107,109,111,113,115,117,119,121,123,125,127\nFlags:                 fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc aperfmperf eagerfpu pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch epb cat_l3 invpcid_single intel_pt ssbd mba ibrs ibpb stibp ibrs_enhanced tpr_shadow vnmi flexpriority ept vpid fsgsbase tsc_adjust bmi1 hle avx2 smep bmi2 erms invpcid rtm cqm rdt_a avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local dtherm ida arat pln pts avx512vbmi umip pku ospke avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg avx512_vpopcntdq md_clear pconfig spec_ctrl intel_stibp flush_l1d arch_capabilities\n\nVersions of relevant libraries:\n[pip3] flake8==7.0.0\n[pip3] mypy==1.10.0\n[pip3] mypy-extensions==1.0.0\n[pip3] numpy==1.26.4\n[pip3] numpydoc==1.7.0\n[pip3] torch==2.4.1\n[pip3] torchaudio==2.4.1\n[pip3] torchvision==0.19.1\n[pip3] triton==3.0.0\n[conda] _anaconda_depends         2024.06             py311_mkl_2  \n[conda] blas                      1.0                         mkl  \n[conda] ffmpeg                    4.3                  hf484d3e_0    pytorch\n[conda] libjpeg-turbo             2.0.0                h9bf148f_0    pytorch\n[conda] mkl                       2023.1.0         h213fc3f_46344  \n[conda] mkl-service               2.4.0           py311h5eee18b_1  \n[conda] mkl_fft                   1.3.10          py311h5eee18b_0  \n[conda] mkl_random                1.2.7           py311ha02d727_0  \n[conda] numpy                     1.26.4          py311h08b1b3b_0  \n[conda] numpy-base                1.26.4          py311hf175353_0  \n[conda] numpydoc                  1.7.0           py311h06a4308_0  \n[conda] pytorch                   2.4.1           py3.11_cuda12.1_cudnn9.1.0_0    pytorch\n[conda] pytorch-cuda              12.1                 ha16c6d3_5    pytorch\n[conda] pytorch-mutex             1.0                        cuda    pytorch\n[conda] torchaudio                2.4.1               py311_cu121    pytorch\n[conda] torchtriton               3.0.0                     py311    pytorch\n[conda] torchvision               0.19.1              py311_cu121    pytorch",
+  "transformers_version": "4.44.2",
+  "upper_git_hash": null
+}