Simon Strandgaard
commited on
Commit
·
437ee94
1
Parent(s):
0f07150
Changes from PlanExe repo
Browse files
src/plan/data/simple_plan_prompts.jsonl
CHANGED
@@ -1,3 +1,4 @@
|
|
|
|
1 |
{"id": "fdbac6bc-6853-47f3-b7ec-bc0051314952", "prompt": "I'm envisioning a streamlined global language—free of archaic features like gendered terms and excessive suffixes, taking cues from LLM tokenization. Some regions might only choose to adopt certain parts of this modern language. Would humanity ultimately benefit more from preserving many distinct languages, or uniting around a single, optimized one?", "tags": ["language", "tokenization"]}
|
2 |
{"id": "762b64e2-5ac8-4684-807a-efd3e81d6bc1", "prompt": "Create a detailed report examining the current situation of microplastics within the world's oceans.", "tags": ["ocean", "microplastics", "climate change", "sustainability"]}
|
3 |
{"id": "930c2abc-faa7-4c21-8ae1-f0323cbcd120", "prompt": "Open the first space elevator terminal in Berlin, Germany, connecting Earths surface to orbit.", "tags": ["space", "exploration", "berlin", "germany"]}
|
|
|
1 |
+
{"id": "f847a181-c9b8-419f-8aef-552e1a3b662f", "prompt": "Distill Arxiv papers into an objective, hype-free summary that indicates whether improvements are truly significant or just noise. Compare claims with benchmarks, flag inflated gains, and foster a clear, evidence-based understanding of machine learning progress without marketing language. To make the distilled data available with minimal upkeep and maximum longevity, publish these summaries as an open-access dataset on a well-established repository.", "tags": ["Arxiv", "paper", "dataset", "signal", "noise"]}
|
2 |
{"id": "fdbac6bc-6853-47f3-b7ec-bc0051314952", "prompt": "I'm envisioning a streamlined global language—free of archaic features like gendered terms and excessive suffixes, taking cues from LLM tokenization. Some regions might only choose to adopt certain parts of this modern language. Would humanity ultimately benefit more from preserving many distinct languages, or uniting around a single, optimized one?", "tags": ["language", "tokenization"]}
|
3 |
{"id": "762b64e2-5ac8-4684-807a-efd3e81d6bc1", "prompt": "Create a detailed report examining the current situation of microplastics within the world's oceans.", "tags": ["ocean", "microplastics", "climate change", "sustainability"]}
|
4 |
{"id": "930c2abc-faa7-4c21-8ae1-f0323cbcd120", "prompt": "Open the first space elevator terminal in Berlin, Germany, connecting Earths surface to orbit.", "tags": ["space", "exploration", "berlin", "germany"]}
|
src/plan/run_plan_pipeline.py
CHANGED
@@ -268,7 +268,8 @@ class SWOTAnalysisTask(PlanTask):
|
|
268 |
json.dump(swot_raw_dict, f, indent=2)
|
269 |
|
270 |
# Write the SWOT analysis as Markdown.
|
271 |
-
|
|
|
272 |
f.write(swot_markdown)
|
273 |
|
274 |
logger.info("SWOT analysis complete.")
|
@@ -310,7 +311,8 @@ class ExpertReviewTask(PlanTask):
|
|
310 |
pre_project_assessment_dict = json.load(f)
|
311 |
with self.input()['project_plan'].open("r") as f:
|
312 |
project_plan_dict = json.load(f)
|
313 |
-
|
|
|
314 |
swot_markdown = f.read()
|
315 |
|
316 |
# Build the query.
|
|
|
268 |
json.dump(swot_raw_dict, f, indent=2)
|
269 |
|
270 |
# Write the SWOT analysis as Markdown.
|
271 |
+
markdown_path = self.output()['markdown'].path
|
272 |
+
with open(markdown_path, "w", encoding="utf-8") as f:
|
273 |
f.write(swot_markdown)
|
274 |
|
275 |
logger.info("SWOT analysis complete.")
|
|
|
311 |
pre_project_assessment_dict = json.load(f)
|
312 |
with self.input()['project_plan'].open("r") as f:
|
313 |
project_plan_dict = json.load(f)
|
314 |
+
swot_markdown_path = self.input()['swot_analysis']['markdown'].path
|
315 |
+
with open(swot_markdown_path, "r", encoding="utf-8") as f:
|
316 |
swot_markdown = f.read()
|
317 |
|
318 |
# Build the query.
|