edsl 0.1.39.dev2__py3-none-any.whl → 0.1.39.dev3__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (334) hide show
  1. edsl/Base.py +332 -385
  2. edsl/BaseDiff.py +260 -260
  3. edsl/TemplateLoader.py +24 -24
  4. edsl/__init__.py +49 -57
  5. edsl/__version__.py +1 -1
  6. edsl/agents/Agent.py +867 -1079
  7. edsl/agents/AgentList.py +413 -551
  8. edsl/agents/Invigilator.py +233 -285
  9. edsl/agents/InvigilatorBase.py +270 -254
  10. edsl/agents/PromptConstructor.py +354 -252
  11. edsl/agents/__init__.py +3 -2
  12. edsl/agents/descriptors.py +99 -99
  13. edsl/agents/prompt_helpers.py +129 -129
  14. edsl/auto/AutoStudy.py +117 -117
  15. edsl/auto/StageBase.py +230 -230
  16. edsl/auto/StageGenerateSurvey.py +178 -178
  17. edsl/auto/StageLabelQuestions.py +125 -125
  18. edsl/auto/StagePersona.py +61 -61
  19. edsl/auto/StagePersonaDimensionValueRanges.py +88 -88
  20. edsl/auto/StagePersonaDimensionValues.py +74 -74
  21. edsl/auto/StagePersonaDimensions.py +69 -69
  22. edsl/auto/StageQuestions.py +73 -73
  23. edsl/auto/SurveyCreatorPipeline.py +21 -21
  24. edsl/auto/utilities.py +224 -224
  25. edsl/base/Base.py +279 -279
  26. edsl/config.py +157 -177
  27. edsl/conversation/Conversation.py +290 -290
  28. edsl/conversation/car_buying.py +58 -59
  29. edsl/conversation/chips.py +95 -95
  30. edsl/conversation/mug_negotiation.py +81 -81
  31. edsl/conversation/next_speaker_utilities.py +93 -93
  32. edsl/coop/PriceFetcher.py +54 -54
  33. edsl/coop/__init__.py +2 -2
  34. edsl/coop/coop.py +1028 -1090
  35. edsl/coop/utils.py +131 -131
  36. edsl/data/Cache.py +555 -562
  37. edsl/data/CacheEntry.py +233 -230
  38. edsl/data/CacheHandler.py +149 -170
  39. edsl/data/RemoteCacheSync.py +78 -78
  40. edsl/data/SQLiteDict.py +292 -292
  41. edsl/data/__init__.py +4 -5
  42. edsl/data/orm.py +10 -10
  43. edsl/data_transfer_models.py +73 -74
  44. edsl/enums.py +175 -195
  45. edsl/exceptions/BaseException.py +21 -21
  46. edsl/exceptions/__init__.py +54 -54
  47. edsl/exceptions/agents.py +42 -54
  48. edsl/exceptions/cache.py +5 -5
  49. edsl/exceptions/configuration.py +16 -16
  50. edsl/exceptions/coop.py +10 -10
  51. edsl/exceptions/data.py +14 -14
  52. edsl/exceptions/general.py +34 -34
  53. edsl/exceptions/jobs.py +33 -33
  54. edsl/exceptions/language_models.py +63 -63
  55. edsl/exceptions/prompts.py +15 -15
  56. edsl/exceptions/questions.py +91 -109
  57. edsl/exceptions/results.py +29 -29
  58. edsl/exceptions/scenarios.py +22 -29
  59. edsl/exceptions/surveys.py +37 -37
  60. edsl/inference_services/AnthropicService.py +87 -84
  61. edsl/inference_services/AwsBedrock.py +120 -118
  62. edsl/inference_services/AzureAI.py +217 -215
  63. edsl/inference_services/DeepInfraService.py +18 -18
  64. edsl/inference_services/GoogleService.py +148 -139
  65. edsl/inference_services/GroqService.py +20 -20
  66. edsl/inference_services/InferenceServiceABC.py +147 -80
  67. edsl/inference_services/InferenceServicesCollection.py +97 -122
  68. edsl/inference_services/MistralAIService.py +123 -120
  69. edsl/inference_services/OllamaService.py +18 -18
  70. edsl/inference_services/OpenAIService.py +224 -221
  71. edsl/inference_services/PerplexityService.py +163 -160
  72. edsl/inference_services/TestService.py +89 -92
  73. edsl/inference_services/TogetherAIService.py +170 -170
  74. edsl/inference_services/models_available_cache.py +118 -118
  75. edsl/inference_services/rate_limits_cache.py +25 -25
  76. edsl/inference_services/registry.py +41 -41
  77. edsl/inference_services/write_available.py +10 -10
  78. edsl/jobs/Answers.py +56 -43
  79. edsl/jobs/Jobs.py +898 -757
  80. edsl/jobs/JobsChecks.py +147 -172
  81. edsl/jobs/JobsPrompts.py +268 -270
  82. edsl/jobs/JobsRemoteInferenceHandler.py +239 -287
  83. edsl/jobs/__init__.py +1 -1
  84. edsl/jobs/buckets/BucketCollection.py +63 -104
  85. edsl/jobs/buckets/ModelBuckets.py +65 -65
  86. edsl/jobs/buckets/TokenBucket.py +251 -283
  87. edsl/jobs/interviews/Interview.py +661 -358
  88. edsl/jobs/interviews/InterviewExceptionCollection.py +99 -99
  89. edsl/jobs/interviews/InterviewExceptionEntry.py +186 -186
  90. edsl/jobs/interviews/InterviewStatistic.py +63 -63
  91. edsl/jobs/interviews/InterviewStatisticsCollection.py +25 -25
  92. edsl/jobs/interviews/InterviewStatusDictionary.py +78 -78
  93. edsl/jobs/interviews/InterviewStatusLog.py +92 -92
  94. edsl/jobs/interviews/ReportErrors.py +66 -66
  95. edsl/jobs/interviews/interview_status_enum.py +9 -9
  96. edsl/jobs/runners/JobsRunnerAsyncio.py +466 -421
  97. edsl/jobs/runners/JobsRunnerStatus.py +330 -330
  98. edsl/jobs/tasks/QuestionTaskCreator.py +242 -244
  99. edsl/jobs/tasks/TaskCreators.py +64 -64
  100. edsl/jobs/tasks/TaskHistory.py +450 -449
  101. edsl/jobs/tasks/TaskStatusLog.py +23 -23
  102. edsl/jobs/tasks/task_status_enum.py +163 -161
  103. edsl/jobs/tokens/InterviewTokenUsage.py +27 -27
  104. edsl/jobs/tokens/TokenUsage.py +34 -34
  105. edsl/language_models/KeyLookup.py +30 -0
  106. edsl/language_models/LanguageModel.py +668 -571
  107. edsl/language_models/ModelList.py +155 -153
  108. edsl/language_models/RegisterLanguageModelsMeta.py +184 -184
  109. edsl/language_models/__init__.py +3 -2
  110. edsl/language_models/fake_openai_call.py +15 -15
  111. edsl/language_models/fake_openai_service.py +61 -61
  112. edsl/language_models/registry.py +190 -180
  113. edsl/language_models/repair.py +156 -156
  114. edsl/language_models/unused/ReplicateBase.py +83 -0
  115. edsl/language_models/utilities.py +64 -65
  116. edsl/notebooks/Notebook.py +258 -263
  117. edsl/notebooks/__init__.py +1 -1
  118. edsl/prompts/Prompt.py +362 -352
  119. edsl/prompts/__init__.py +2 -2
  120. edsl/questions/AnswerValidatorMixin.py +289 -334
  121. edsl/questions/QuestionBase.py +664 -509
  122. edsl/questions/QuestionBaseGenMixin.py +161 -165
  123. edsl/questions/QuestionBasePromptsMixin.py +217 -221
  124. edsl/questions/QuestionBudget.py +227 -227
  125. edsl/questions/QuestionCheckBox.py +359 -359
  126. edsl/questions/QuestionExtract.py +182 -182
  127. edsl/questions/QuestionFreeText.py +114 -113
  128. edsl/questions/QuestionFunctional.py +166 -166
  129. edsl/questions/QuestionList.py +231 -229
  130. edsl/questions/QuestionMultipleChoice.py +286 -330
  131. edsl/questions/QuestionNumerical.py +153 -151
  132. edsl/questions/QuestionRank.py +324 -314
  133. edsl/questions/Quick.py +41 -41
  134. edsl/questions/RegisterQuestionsMeta.py +71 -71
  135. edsl/questions/ResponseValidatorABC.py +174 -200
  136. edsl/questions/SimpleAskMixin.py +73 -74
  137. edsl/questions/__init__.py +26 -27
  138. edsl/questions/compose_questions.py +98 -98
  139. edsl/questions/decorators.py +21 -21
  140. edsl/questions/derived/QuestionLikertFive.py +76 -76
  141. edsl/questions/derived/QuestionLinearScale.py +87 -90
  142. edsl/questions/derived/QuestionTopK.py +93 -93
  143. edsl/questions/derived/QuestionYesNo.py +82 -82
  144. edsl/questions/descriptors.py +413 -427
  145. edsl/questions/prompt_templates/question_budget.jinja +13 -13
  146. edsl/questions/prompt_templates/question_checkbox.jinja +32 -32
  147. edsl/questions/prompt_templates/question_extract.jinja +11 -11
  148. edsl/questions/prompt_templates/question_free_text.jinja +3 -3
  149. edsl/questions/prompt_templates/question_linear_scale.jinja +11 -11
  150. edsl/questions/prompt_templates/question_list.jinja +17 -17
  151. edsl/questions/prompt_templates/question_multiple_choice.jinja +33 -33
  152. edsl/questions/prompt_templates/question_numerical.jinja +36 -36
  153. edsl/questions/question_registry.py +177 -177
  154. edsl/questions/settings.py +12 -12
  155. edsl/questions/templates/budget/answering_instructions.jinja +7 -7
  156. edsl/questions/templates/budget/question_presentation.jinja +7 -7
  157. edsl/questions/templates/checkbox/answering_instructions.jinja +10 -10
  158. edsl/questions/templates/checkbox/question_presentation.jinja +22 -22
  159. edsl/questions/templates/extract/answering_instructions.jinja +7 -7
  160. edsl/questions/templates/likert_five/answering_instructions.jinja +10 -10
  161. edsl/questions/templates/likert_five/question_presentation.jinja +11 -11
  162. edsl/questions/templates/linear_scale/answering_instructions.jinja +5 -5
  163. edsl/questions/templates/linear_scale/question_presentation.jinja +5 -5
  164. edsl/questions/templates/list/answering_instructions.jinja +3 -3
  165. edsl/questions/templates/list/question_presentation.jinja +5 -5
  166. edsl/questions/templates/multiple_choice/answering_instructions.jinja +9 -9
  167. edsl/questions/templates/multiple_choice/question_presentation.jinja +11 -11
  168. edsl/questions/templates/numerical/answering_instructions.jinja +6 -6
  169. edsl/questions/templates/numerical/question_presentation.jinja +6 -6
  170. edsl/questions/templates/rank/answering_instructions.jinja +11 -11
  171. edsl/questions/templates/rank/question_presentation.jinja +15 -15
  172. edsl/questions/templates/top_k/answering_instructions.jinja +8 -8
  173. edsl/questions/templates/top_k/question_presentation.jinja +22 -22
  174. edsl/questions/templates/yes_no/answering_instructions.jinja +6 -6
  175. edsl/questions/templates/yes_no/question_presentation.jinja +11 -11
  176. edsl/results/CSSParameterizer.py +108 -108
  177. edsl/results/Dataset.py +424 -587
  178. edsl/results/DatasetExportMixin.py +731 -653
  179. edsl/results/DatasetTree.py +275 -295
  180. edsl/results/Result.py +465 -451
  181. edsl/results/Results.py +1165 -1172
  182. edsl/results/ResultsDBMixin.py +238 -0
  183. edsl/results/ResultsExportMixin.py +43 -45
  184. edsl/results/ResultsFetchMixin.py +33 -33
  185. edsl/results/ResultsGGMixin.py +121 -121
  186. edsl/results/ResultsToolsMixin.py +98 -98
  187. edsl/results/Selector.py +135 -145
  188. edsl/results/TableDisplay.py +198 -125
  189. edsl/results/__init__.py +2 -2
  190. edsl/results/table_display.css +77 -77
  191. edsl/results/tree_explore.py +115 -115
  192. edsl/scenarios/FileStore.py +632 -511
  193. edsl/scenarios/Scenario.py +601 -498
  194. edsl/scenarios/ScenarioHtmlMixin.py +64 -65
  195. edsl/scenarios/ScenarioJoin.py +127 -131
  196. edsl/scenarios/ScenarioList.py +1287 -1430
  197. edsl/scenarios/ScenarioListExportMixin.py +52 -45
  198. edsl/scenarios/ScenarioListPdfMixin.py +261 -239
  199. edsl/scenarios/__init__.py +4 -3
  200. edsl/shared.py +1 -1
  201. edsl/study/ObjectEntry.py +173 -173
  202. edsl/study/ProofOfWork.py +113 -113
  203. edsl/study/SnapShot.py +80 -80
  204. edsl/study/Study.py +528 -521
  205. edsl/study/__init__.py +4 -4
  206. edsl/surveys/DAG.py +148 -148
  207. edsl/surveys/Memory.py +31 -31
  208. edsl/surveys/MemoryPlan.py +244 -244
  209. edsl/surveys/Rule.py +326 -327
  210. edsl/surveys/RuleCollection.py +387 -385
  211. edsl/surveys/Survey.py +1801 -1229
  212. edsl/surveys/SurveyCSS.py +261 -273
  213. edsl/surveys/SurveyExportMixin.py +259 -259
  214. edsl/surveys/{SurveyFlowVisualization.py → SurveyFlowVisualizationMixin.py} +179 -181
  215. edsl/surveys/SurveyQualtricsImport.py +284 -284
  216. edsl/surveys/__init__.py +3 -5
  217. edsl/surveys/base.py +53 -53
  218. edsl/surveys/descriptors.py +56 -60
  219. edsl/surveys/instructions/ChangeInstruction.py +49 -48
  220. edsl/surveys/instructions/Instruction.py +65 -56
  221. edsl/surveys/instructions/InstructionCollection.py +77 -82
  222. edsl/templates/error_reporting/base.html +23 -23
  223. edsl/templates/error_reporting/exceptions_by_model.html +34 -34
  224. edsl/templates/error_reporting/exceptions_by_question_name.html +16 -16
  225. edsl/templates/error_reporting/exceptions_by_type.html +16 -16
  226. edsl/templates/error_reporting/interview_details.html +115 -115
  227. edsl/templates/error_reporting/interviews.html +19 -19
  228. edsl/templates/error_reporting/overview.html +4 -4
  229. edsl/templates/error_reporting/performance_plot.html +1 -1
  230. edsl/templates/error_reporting/report.css +73 -73
  231. edsl/templates/error_reporting/report.html +117 -117
  232. edsl/templates/error_reporting/report.js +25 -25
  233. edsl/tools/__init__.py +1 -1
  234. edsl/tools/clusters.py +192 -192
  235. edsl/tools/embeddings.py +27 -27
  236. edsl/tools/embeddings_plotting.py +118 -118
  237. edsl/tools/plotting.py +112 -112
  238. edsl/tools/summarize.py +18 -18
  239. edsl/utilities/SystemInfo.py +28 -28
  240. edsl/utilities/__init__.py +22 -22
  241. edsl/utilities/ast_utilities.py +25 -25
  242. edsl/utilities/data/Registry.py +6 -6
  243. edsl/utilities/data/__init__.py +1 -1
  244. edsl/utilities/data/scooter_results.json +1 -1
  245. edsl/utilities/decorators.py +77 -77
  246. edsl/utilities/gcp_bucket/cloud_storage.py +96 -96
  247. edsl/utilities/interface.py +627 -627
  248. edsl/utilities/naming_utilities.py +263 -263
  249. edsl/utilities/repair_functions.py +28 -28
  250. edsl/utilities/restricted_python.py +70 -70
  251. edsl/utilities/utilities.py +424 -436
  252. {edsl-0.1.39.dev2.dist-info → edsl-0.1.39.dev3.dist-info}/LICENSE +21 -21
  253. {edsl-0.1.39.dev2.dist-info → edsl-0.1.39.dev3.dist-info}/METADATA +10 -12
  254. edsl-0.1.39.dev3.dist-info/RECORD +277 -0
  255. edsl/agents/QuestionInstructionPromptBuilder.py +0 -128
  256. edsl/agents/QuestionOptionProcessor.py +0 -172
  257. edsl/agents/QuestionTemplateReplacementsBuilder.py +0 -137
  258. edsl/coop/CoopFunctionsMixin.py +0 -15
  259. edsl/coop/ExpectedParrotKeyHandler.py +0 -125
  260. edsl/exceptions/inference_services.py +0 -5
  261. edsl/inference_services/AvailableModelCacheHandler.py +0 -184
  262. edsl/inference_services/AvailableModelFetcher.py +0 -209
  263. edsl/inference_services/ServiceAvailability.py +0 -135
  264. edsl/inference_services/data_structures.py +0 -62
  265. edsl/jobs/AnswerQuestionFunctionConstructor.py +0 -188
  266. edsl/jobs/FetchInvigilator.py +0 -40
  267. edsl/jobs/InterviewTaskManager.py +0 -98
  268. edsl/jobs/InterviewsConstructor.py +0 -48
  269. edsl/jobs/JobsComponentConstructor.py +0 -189
  270. edsl/jobs/JobsRemoteInferenceLogger.py +0 -239
  271. edsl/jobs/RequestTokenEstimator.py +0 -30
  272. edsl/jobs/buckets/TokenBucketAPI.py +0 -211
  273. edsl/jobs/buckets/TokenBucketClient.py +0 -191
  274. edsl/jobs/decorators.py +0 -35
  275. edsl/jobs/jobs_status_enums.py +0 -9
  276. edsl/jobs/loggers/HTMLTableJobLogger.py +0 -304
  277. edsl/language_models/ComputeCost.py +0 -63
  278. edsl/language_models/PriceManager.py +0 -127
  279. edsl/language_models/RawResponseHandler.py +0 -106
  280. edsl/language_models/ServiceDataSources.py +0 -0
  281. edsl/language_models/key_management/KeyLookup.py +0 -63
  282. edsl/language_models/key_management/KeyLookupBuilder.py +0 -273
  283. edsl/language_models/key_management/KeyLookupCollection.py +0 -38
  284. edsl/language_models/key_management/__init__.py +0 -0
  285. edsl/language_models/key_management/models.py +0 -131
  286. edsl/notebooks/NotebookToLaTeX.py +0 -142
  287. edsl/questions/ExceptionExplainer.py +0 -77
  288. edsl/questions/HTMLQuestion.py +0 -103
  289. edsl/questions/LoopProcessor.py +0 -149
  290. edsl/questions/QuestionMatrix.py +0 -265
  291. edsl/questions/ResponseValidatorFactory.py +0 -28
  292. edsl/questions/templates/matrix/__init__.py +0 -1
  293. edsl/questions/templates/matrix/answering_instructions.jinja +0 -5
  294. edsl/questions/templates/matrix/question_presentation.jinja +0 -20
  295. edsl/results/MarkdownToDocx.py +0 -122
  296. edsl/results/MarkdownToPDF.py +0 -111
  297. edsl/results/TextEditor.py +0 -50
  298. edsl/results/smart_objects.py +0 -96
  299. edsl/results/table_data_class.py +0 -12
  300. edsl/results/table_renderers.py +0 -118
  301. edsl/scenarios/ConstructDownloadLink.py +0 -109
  302. edsl/scenarios/DirectoryScanner.py +0 -96
  303. edsl/scenarios/DocumentChunker.py +0 -102
  304. edsl/scenarios/DocxScenario.py +0 -16
  305. edsl/scenarios/PdfExtractor.py +0 -40
  306. edsl/scenarios/ScenarioSelector.py +0 -156
  307. edsl/scenarios/file_methods.py +0 -85
  308. edsl/scenarios/handlers/__init__.py +0 -13
  309. edsl/scenarios/handlers/csv.py +0 -38
  310. edsl/scenarios/handlers/docx.py +0 -76
  311. edsl/scenarios/handlers/html.py +0 -37
  312. edsl/scenarios/handlers/json.py +0 -111
  313. edsl/scenarios/handlers/latex.py +0 -5
  314. edsl/scenarios/handlers/md.py +0 -51
  315. edsl/scenarios/handlers/pdf.py +0 -68
  316. edsl/scenarios/handlers/png.py +0 -39
  317. edsl/scenarios/handlers/pptx.py +0 -105
  318. edsl/scenarios/handlers/py.py +0 -294
  319. edsl/scenarios/handlers/sql.py +0 -313
  320. edsl/scenarios/handlers/sqlite.py +0 -149
  321. edsl/scenarios/handlers/txt.py +0 -33
  322. edsl/surveys/ConstructDAG.py +0 -92
  323. edsl/surveys/EditSurvey.py +0 -221
  324. edsl/surveys/InstructionHandler.py +0 -100
  325. edsl/surveys/MemoryManagement.py +0 -72
  326. edsl/surveys/RuleManager.py +0 -172
  327. edsl/surveys/Simulator.py +0 -75
  328. edsl/surveys/SurveyToApp.py +0 -141
  329. edsl/utilities/PrettyList.py +0 -56
  330. edsl/utilities/is_notebook.py +0 -18
  331. edsl/utilities/is_valid_variable_name.py +0 -11
  332. edsl/utilities/remove_edsl_version.py +0 -24
  333. edsl-0.1.39.dev2.dist-info/RECORD +0 -352
  334. {edsl-0.1.39.dev2.dist-info → edsl-0.1.39.dev3.dist-info}/WHEEL +0 -0
@@ -1,118 +1,118 @@
1
- import json
2
- import numpy as np
3
- from sklearn.manifold import TSNE
4
- from IPython.display import display_html
5
-
6
-
7
- def compute_tsne(embeddings, labels):
8
- embeddings_np = np.array(embeddings)
9
- tsne = TSNE(n_components=2, random_state=42)
10
- tsne_results = tsne.fit_transform(embeddings_np)
11
- data = [
12
- {
13
- "x": float(tsne_results[i, 0]),
14
- "y": float(tsne_results[i, 1]),
15
- "label": labels[i],
16
- }
17
- for i in range(len(labels))
18
- ]
19
- return data
20
-
21
-
22
- def plot_tsne_in_notebook(embeddings, labels):
23
- # Compute t-SNE
24
- data = compute_tsne(embeddings, labels)
25
-
26
- # Convert data to JSON
27
- data_json = json.dumps(data)
28
-
29
- # HTML content with embedded data
30
- html_content = f"""
31
- <!DOCTYPE html>
32
- <html lang="en">
33
- <head>
34
- <meta charset="UTF-8">
35
- <title>t-SNE Plot with D3.js</title>
36
- <script src="https://d3js.org/d3.v6.min.js"></script>
37
- <style>
38
- .tooltip {{
39
- position: absolute;
40
- text-align: center;
41
- width: 80px;
42
- height: 28px;
43
- padding: 2px;
44
- font: 12px sans-serif;
45
- background: lightsteelblue;
46
- border: 0px;
47
- border-radius: 8px;
48
- pointer-events: none;
49
- }}
50
- </style>
51
- </head>
52
- <body>
53
- <svg width="600" height="600"></svg>
54
-
55
- <script>
56
- // Embedded data
57
- const data = {data_json};
58
-
59
- const svg = d3.select("svg"),
60
- width = +svg.attr("width"),
61
- height = +svg.attr("height");
62
-
63
- // Set up scales
64
- const x = d3.scaleLinear()
65
- .domain(d3.extent(data, d => d.x))
66
- .range([0, width]);
67
-
68
- const y = d3.scaleLinear()
69
- .domain(d3.extent(data, d => d.y))
70
- .range([height, 0]);
71
-
72
- // Create tooltip
73
- const tooltip = d3.select("body").append("div")
74
- .attr("class", "tooltip")
75
- .style("opacity", 0);
76
-
77
- // Create circles for each point
78
- svg.selectAll("circle")
79
- .data(data)
80
- .enter().append("circle")
81
- .attr("cx", d => x(d.x))
82
- .attr("cy", d => y(d.y))
83
- .attr("r", 5)
84
- .style("fill", "steelblue")
85
- .on("mouseover", function(event, d) {{
86
- tooltip.transition()
87
- .duration(200)
88
- .style("opacity", .9);
89
- tooltip.html(d.label)
90
- .style("left", (event.pageX + 5) + "px")
91
- .style("top", (event.pageY - 28) + "px");
92
- }})
93
- .on("mouseout", function(d) {{
94
- tooltip.transition()
95
- .duration(500)
96
- .style("opacity", 0);
97
- }});
98
- </script>
99
- </body>
100
- </html>
101
- """
102
-
103
- # Write HTML content to a temporary file
104
- html_file = "tsne_plot.html"
105
- with open(html_file, "w") as file:
106
- file.write(html_content)
107
-
108
- # Display the HTML content in an iframe within a Jupyter notebook
109
- display_html(
110
- f'<iframe src="{html_file}" width="600" height="600"></iframe>', raw=True
111
- )
112
-
113
-
114
- # Example usage
115
- if __name__ == "__main__":
116
- embeddings = [[0.1, 0.2, 0.3], [0.4, 0.5, 0.6], [0.7, 0.8, 0.9]]
117
- labels = ["String 1", "String 2", "String 3"]
118
- plot_tsne_in_notebook(embeddings, labels)
1
+ import json
2
+ import numpy as np
3
+ from sklearn.manifold import TSNE
4
+ from IPython.display import display_html
5
+
6
+
7
+ def compute_tsne(embeddings, labels):
8
+ embeddings_np = np.array(embeddings)
9
+ tsne = TSNE(n_components=2, random_state=42)
10
+ tsne_results = tsne.fit_transform(embeddings_np)
11
+ data = [
12
+ {
13
+ "x": float(tsne_results[i, 0]),
14
+ "y": float(tsne_results[i, 1]),
15
+ "label": labels[i],
16
+ }
17
+ for i in range(len(labels))
18
+ ]
19
+ return data
20
+
21
+
22
+ def plot_tsne_in_notebook(embeddings, labels):
23
+ # Compute t-SNE
24
+ data = compute_tsne(embeddings, labels)
25
+
26
+ # Convert data to JSON
27
+ data_json = json.dumps(data)
28
+
29
+ # HTML content with embedded data
30
+ html_content = f"""
31
+ <!DOCTYPE html>
32
+ <html lang="en">
33
+ <head>
34
+ <meta charset="UTF-8">
35
+ <title>t-SNE Plot with D3.js</title>
36
+ <script src="https://d3js.org/d3.v6.min.js"></script>
37
+ <style>
38
+ .tooltip {{
39
+ position: absolute;
40
+ text-align: center;
41
+ width: 80px;
42
+ height: 28px;
43
+ padding: 2px;
44
+ font: 12px sans-serif;
45
+ background: lightsteelblue;
46
+ border: 0px;
47
+ border-radius: 8px;
48
+ pointer-events: none;
49
+ }}
50
+ </style>
51
+ </head>
52
+ <body>
53
+ <svg width="600" height="600"></svg>
54
+
55
+ <script>
56
+ // Embedded data
57
+ const data = {data_json};
58
+
59
+ const svg = d3.select("svg"),
60
+ width = +svg.attr("width"),
61
+ height = +svg.attr("height");
62
+
63
+ // Set up scales
64
+ const x = d3.scaleLinear()
65
+ .domain(d3.extent(data, d => d.x))
66
+ .range([0, width]);
67
+
68
+ const y = d3.scaleLinear()
69
+ .domain(d3.extent(data, d => d.y))
70
+ .range([height, 0]);
71
+
72
+ // Create tooltip
73
+ const tooltip = d3.select("body").append("div")
74
+ .attr("class", "tooltip")
75
+ .style("opacity", 0);
76
+
77
+ // Create circles for each point
78
+ svg.selectAll("circle")
79
+ .data(data)
80
+ .enter().append("circle")
81
+ .attr("cx", d => x(d.x))
82
+ .attr("cy", d => y(d.y))
83
+ .attr("r", 5)
84
+ .style("fill", "steelblue")
85
+ .on("mouseover", function(event, d) {{
86
+ tooltip.transition()
87
+ .duration(200)
88
+ .style("opacity", .9);
89
+ tooltip.html(d.label)
90
+ .style("left", (event.pageX + 5) + "px")
91
+ .style("top", (event.pageY - 28) + "px");
92
+ }})
93
+ .on("mouseout", function(d) {{
94
+ tooltip.transition()
95
+ .duration(500)
96
+ .style("opacity", 0);
97
+ }});
98
+ </script>
99
+ </body>
100
+ </html>
101
+ """
102
+
103
+ # Write HTML content to a temporary file
104
+ html_file = "tsne_plot.html"
105
+ with open(html_file, "w") as file:
106
+ file.write(html_content)
107
+
108
+ # Display the HTML content in an iframe within a Jupyter notebook
109
+ display_html(
110
+ f'<iframe src="{html_file}" width="600" height="600"></iframe>', raw=True
111
+ )
112
+
113
+
114
+ # Example usage
115
+ if __name__ == "__main__":
116
+ embeddings = [[0.1, 0.2, 0.3], [0.4, 0.5, 0.6], [0.7, 0.8, 0.9]]
117
+ labels = ["String 1", "String 2", "String 3"]
118
+ plot_tsne_in_notebook(embeddings, labels)
edsl/tools/plotting.py CHANGED
@@ -1,112 +1,112 @@
1
- from typing import Optional
2
-
3
-
4
- def count_query(field):
5
- return f"""SELECT
6
- {field},
7
- COUNT(*) as number
8
- FROM self
9
- GROUP BY {field}
10
- """
11
-
12
-
13
- def get_options(results, field):
14
- question_type = results.survey._get_question_by_name(field).question_type
15
- if question_type in ["multiple_choice", "checkbox"]:
16
- return results.select(f"{field}_question_options").first()
17
- else:
18
- return None
19
-
20
-
21
- def interpret_image(path, analysis):
22
- from edsl import QuestionFreeText
23
- from edsl import Model
24
- from edsl import Scenario
25
-
26
- s = Scenario.from_image(path)
27
- if isinstance(analysis, str):
28
- plot_question_texts = [analysis]
29
- elif isinstance(analysis, list):
30
- plot_question_texts = analysis
31
-
32
- scenario_list = s.replicate(len(plot_question_texts))
33
- scenario_list.add_list("plot_question_text", plot_question_texts)
34
-
35
- m = Model("gpt-4o")
36
- q = QuestionFreeText(
37
- question_text="{{ plot_question_text }}", question_name="interpretation"
38
- )
39
- results = q.by(m).by(scenario_list).run()
40
- return results.select("plot_question_text", "interpretation").print(
41
- format="rich",
42
- pretty_labels={
43
- "scenario.plot_question_text": "Question to the model",
44
- "answer.interpretation": "Model answer",
45
- },
46
- )
47
-
48
-
49
- def barchart(
50
- results,
51
- field: str,
52
- fetch_options=True,
53
- xlab: Optional[str] = None,
54
- ylab: Optional[str] = None,
55
- analysis: Optional[str] = None,
56
- format: str = "png",
57
- ):
58
- labels = ""
59
- if xlab:
60
- labels += f"+ xlab('{xlab}')"
61
- if ylab:
62
- labels += f"+ ylab('{ylab}')"
63
-
64
- if fetch_options:
65
- factor_orders = {field: get_options(results, field)}
66
- else:
67
- factor_orders = None
68
-
69
- plot = results.ggplot2(
70
- f"""ggplot(data = self, aes(x = {field}, y = number)) +
71
- geom_bar(stat = "identity") +
72
- theme_bw() +
73
- theme(axis.text.x = element_text(angle = 45, hjust = 1)) {labels}""",
74
- sql=count_query(field),
75
- factor_orders=factor_orders,
76
- format=format,
77
- filename=f"barchart_{field}.{format}",
78
- )
79
- if analysis:
80
- interpret_image(f"barchart_{field}.{format}", analysis)
81
-
82
- return plot
83
-
84
-
85
- def theme_plot(results, field, context, themes=None, progress_bar=False):
86
- _, themes = results.auto_theme(
87
- field=field, context=context, themes=themes, progress_bar=progress_bar
88
- )
89
-
90
- themes_query = f"""
91
- SELECT theme, COUNT(*) AS mentions
92
- FROM (
93
- SELECT json_each.value AS theme
94
- FROM self,
95
- json_each({ field }_themes)
96
- )
97
- GROUP BY theme
98
- HAVING theme <> 'Other'
99
- ORDER BY mentions DESC
100
- """
101
- themes = results.sql(themes_query, to_list=True)
102
-
103
- (
104
- results.filter(f"{field} != ''").ggplot2(
105
- """ggplot(data = self, aes(x = theme, y = mentions)) +
106
- geom_bar(stat = "identity") +
107
- coord_flip() +
108
- theme_bw()""",
109
- sql=themes_query,
110
- factor_orders={"theme": [t[0] for t in themes]},
111
- )
112
- )
1
+ from typing import Optional
2
+
3
+
4
+ def count_query(field):
5
+ return f"""SELECT
6
+ {field},
7
+ COUNT(*) as number
8
+ FROM self
9
+ GROUP BY {field}
10
+ """
11
+
12
+
13
+ def get_options(results, field):
14
+ question_type = results.survey.get_question(field).question_type
15
+ if question_type in ["multiple_choice", "checkbox"]:
16
+ return results.select(f"{field}_question_options").first()
17
+ else:
18
+ return None
19
+
20
+
21
+ def interpret_image(path, analysis):
22
+ from edsl import QuestionFreeText
23
+ from edsl import Model
24
+ from edsl import Scenario
25
+
26
+ s = Scenario.from_image(path)
27
+ if isinstance(analysis, str):
28
+ plot_question_texts = [analysis]
29
+ elif isinstance(analysis, list):
30
+ plot_question_texts = analysis
31
+
32
+ scenario_list = s.replicate(len(plot_question_texts))
33
+ scenario_list.add_list("plot_question_text", plot_question_texts)
34
+
35
+ m = Model("gpt-4o")
36
+ q = QuestionFreeText(
37
+ question_text="{{ plot_question_text }}", question_name="interpretation"
38
+ )
39
+ results = q.by(m).by(scenario_list).run()
40
+ return results.select("plot_question_text", "interpretation").print(
41
+ format="rich",
42
+ pretty_labels={
43
+ "scenario.plot_question_text": "Question to the model",
44
+ "answer.interpretation": "Model answer",
45
+ },
46
+ )
47
+
48
+
49
+ def barchart(
50
+ results,
51
+ field: str,
52
+ fetch_options=True,
53
+ xlab: Optional[str] = None,
54
+ ylab: Optional[str] = None,
55
+ analysis: Optional[str] = None,
56
+ format: str = "png",
57
+ ):
58
+ labels = ""
59
+ if xlab:
60
+ labels += f"+ xlab('{xlab}')"
61
+ if ylab:
62
+ labels += f"+ ylab('{ylab}')"
63
+
64
+ if fetch_options:
65
+ factor_orders = {field: get_options(results, field)}
66
+ else:
67
+ factor_orders = None
68
+
69
+ plot = results.ggplot2(
70
+ f"""ggplot(data = self, aes(x = {field}, y = number)) +
71
+ geom_bar(stat = "identity") +
72
+ theme_bw() +
73
+ theme(axis.text.x = element_text(angle = 45, hjust = 1)) {labels}""",
74
+ sql=count_query(field),
75
+ factor_orders=factor_orders,
76
+ format=format,
77
+ filename=f"barchart_{field}.{format}",
78
+ )
79
+ if analysis:
80
+ interpret_image(f"barchart_{field}.{format}", analysis)
81
+
82
+ return plot
83
+
84
+
85
+ def theme_plot(results, field, context, themes=None, progress_bar=False):
86
+ _, themes = results.auto_theme(
87
+ field=field, context=context, themes=themes, progress_bar=progress_bar
88
+ )
89
+
90
+ themes_query = f"""
91
+ SELECT theme, COUNT(*) AS mentions
92
+ FROM (
93
+ SELECT json_each.value AS theme
94
+ FROM self,
95
+ json_each({ field }_themes)
96
+ )
97
+ GROUP BY theme
98
+ HAVING theme <> 'Other'
99
+ ORDER BY mentions DESC
100
+ """
101
+ themes = results.sql(themes_query, to_list=True)
102
+
103
+ (
104
+ results.filter(f"{field} != ''").ggplot2(
105
+ """ggplot(data = self, aes(x = theme, y = mentions)) +
106
+ geom_bar(stat = "identity") +
107
+ coord_flip() +
108
+ theme_bw()""",
109
+ sql=themes_query,
110
+ factor_orders={"theme": [t[0] for t in themes]},
111
+ )
112
+ )
edsl/tools/summarize.py CHANGED
@@ -1,18 +1,18 @@
1
- from edsl import QuestionList, Scenario, Model
2
-
3
-
4
- def summarize(texts, seed_phrase, n_bullets, n_words, models=None):
5
- if models is None:
6
- models = Model()
7
- s = Scenario(
8
- text=texts, seed_phrase=seed_phrase, n_bullets=n_bullets, n_words=n_words
9
- ).expand("text")
10
- QuestionList(
11
- question_text="""
12
- I have the following TEXT EXAMPLE :
13
- {{ text_example_json }}
14
- Please summarize the main point of this EXAMPLE {{seed_phrase }} into {{ n_bullets }} bullet points, where
15
- each bullet point is a {{ n_words }} word phrase.
16
- """,
17
- question_name="summarize",
18
- ).by(s).by(models).run()
1
+ from edsl import QuestionList, Scenario, Model
2
+
3
+
4
+ def summarize(texts, seed_phrase, n_bullets, n_words, models=None):
5
+ if models is None:
6
+ models = Model()
7
+ s = Scenario(
8
+ text=texts, seed_phrase=seed_phrase, n_bullets=n_bullets, n_words=n_words
9
+ ).expand("text")
10
+ QuestionList(
11
+ question_text="""
12
+ I have the following TEXT EXAMPLE :
13
+ {{ text_example_json }}
14
+ Please summarize the main point of this EXAMPLE {{seed_phrase }} into {{ n_bullets }} bullet points, where
15
+ each bullet point is a {{ n_words }} word phrase.
16
+ """,
17
+ question_name="summarize",
18
+ ).by(s).by(models).run()
@@ -1,28 +1,28 @@
1
- """Module to store system information."""
2
-
3
- from dataclasses import dataclass
4
- import getpass
5
- import platform
6
- import pkg_resources
7
-
8
-
9
- @dataclass
10
- class SystemInfo:
11
- """Dataclass to store system information."""
12
-
13
- username: str
14
- system_info: str
15
- release_info: str
16
- package_name: str
17
- package_version: str
18
-
19
- def __init__(self, package_name: str):
20
- """Initialize the dataclass with system."""
21
- self.username = getpass.getuser()
22
- self.system_info = platform.system()
23
- self.release_info = platform.release()
24
- self.package_name = package_name
25
- try:
26
- self.package_version = pkg_resources.get_distribution(package_name).version
27
- except pkg_resources.DistributionNotFound:
28
- self.package_version = "Not installed"
1
+ """Module to store system information."""
2
+
3
+ from dataclasses import dataclass
4
+ import getpass
5
+ import platform
6
+ import pkg_resources
7
+
8
+
9
+ @dataclass
10
+ class SystemInfo:
11
+ """Dataclass to store system information."""
12
+
13
+ username: str
14
+ system_info: str
15
+ release_info: str
16
+ package_name: str
17
+ package_version: str
18
+
19
+ def __init__(self, package_name: str):
20
+ """Initialize the dataclass with system."""
21
+ self.username = getpass.getuser()
22
+ self.system_info = platform.system()
23
+ self.release_info = platform.release()
24
+ self.package_name = package_name
25
+ try:
26
+ self.package_version = pkg_resources.get_distribution(package_name).version
27
+ except pkg_resources.DistributionNotFound:
28
+ self.package_version = "Not installed"
@@ -1,22 +1,22 @@
1
- # from edsl.utilities.interface import (
2
- # print_dict_as_html_table,
3
- # print_dict_with_rich,
4
- # print_list_of_dicts_as_html_table,
5
- # print_table_with_rich,
6
- # print_public_methods_with_doc,
7
- # print_list_of_dicts_as_markdown_table,
8
- # )
9
-
10
- # from edsl.utilities.utilities import (
11
- # create_valid_var_name,
12
- # dict_to_html,
13
- # hash_value,
14
- # HTMLSnippet,
15
- # is_notebook,
16
- # is_gzipped,
17
- # is_valid_variable_name,
18
- # random_string,
19
- # repair_json,
20
- # shorten_string,
21
- # time_all_functions,
22
- # )
1
+ # from edsl.utilities.interface import (
2
+ # print_dict_as_html_table,
3
+ # print_dict_with_rich,
4
+ # print_list_of_dicts_as_html_table,
5
+ # print_table_with_rich,
6
+ # print_public_methods_with_doc,
7
+ # print_list_of_dicts_as_markdown_table,
8
+ # )
9
+
10
+ # from edsl.utilities.utilities import (
11
+ # create_valid_var_name,
12
+ # dict_to_html,
13
+ # hash_value,
14
+ # HTMLSnippet,
15
+ # is_notebook,
16
+ # is_gzipped,
17
+ # is_valid_variable_name,
18
+ # random_string,
19
+ # repair_json,
20
+ # shorten_string,
21
+ # time_all_functions,
22
+ # )
@@ -1,25 +1,25 @@
1
- """Utilities for working with abstract syntax trees (ASTs)."""
2
-
3
- import ast
4
-
5
-
6
- def extract_variable_names(node):
7
- """Extract variable names from an abstract syntax tree (AST) node."""
8
- if isinstance(node, ast.Name):
9
- return [node.id] # Extract variable name
10
- elif isinstance(node, ast.BinOp):
11
- left_names = extract_variable_names(node.left)
12
- right_names = extract_variable_names(node.right)
13
- return left_names + right_names
14
- elif isinstance(node, ast.UnaryOp):
15
- return extract_variable_names(node.operand)
16
- elif isinstance(node, ast.Call):
17
- names = []
18
- for arg in node.args:
19
- names.extend(extract_variable_names(arg))
20
- return names
21
- else:
22
- names = []
23
- for child in ast.iter_child_nodes(node):
24
- names.extend(extract_variable_names(child))
25
- return names
1
+ """Utilities for working with abstract syntax trees (ASTs)."""
2
+
3
+ import ast
4
+
5
+
6
+ def extract_variable_names(node):
7
+ """Extract variable names from an abstract syntax tree (AST) node."""
8
+ if isinstance(node, ast.Name):
9
+ return [node.id] # Extract variable name
10
+ elif isinstance(node, ast.BinOp):
11
+ left_names = extract_variable_names(node.left)
12
+ right_names = extract_variable_names(node.right)
13
+ return left_names + right_names
14
+ elif isinstance(node, ast.UnaryOp):
15
+ return extract_variable_names(node.operand)
16
+ elif isinstance(node, ast.Call):
17
+ names = []
18
+ for arg in node.args:
19
+ names.extend(extract_variable_names(arg))
20
+ return names
21
+ else:
22
+ names = []
23
+ for child in ast.iter_child_nodes(node):
24
+ names.extend(extract_variable_names(child))
25
+ return names
@@ -1,6 +1,6 @@
1
- """This file contains the path to a results file."""
2
-
3
- import os
4
- from edsl import BASE_DIR
5
-
6
- SCOOTER_RESULTS_PATH = os.path.join(BASE_DIR, "utilities/data/scooter_results.json")
1
+ """This file contains the path to a results file."""
2
+
3
+ import os
4
+ from edsl import BASE_DIR
5
+
6
+ SCOOTER_RESULTS_PATH = os.path.join(BASE_DIR, "utilities/data/scooter_results.json")
@@ -1 +1 @@
1
- from edsl.utilities.data.Registry import SCOOTER_RESULTS_PATH
1
+ from edsl.utilities.data.Registry import SCOOTER_RESULTS_PATH