Datasets:
schema_version int64 | batch string | profile string | capability string | capability_ids list | scenario_ledger dict | text string | baseline_checkpoint string | teacher_model string | reviewer_model string | review unknown | token_count int64 | capability_pack_sha256 string | author_model string | source_family string | source_model string | source_path string | source_line int64 | source_split string | curriculum_generation string | compiled_source_artifact_path string | compiled_source_artifact_sha256 string | compiled_source_artifact_line int64 | sample_weight int64 | judge_model string | judge_is_independent bool | independent_review dict | teacher_response_content string | teacher_response dict | teacher_prompt_sha256 string | conversation_sha256 string | orchestration_seed int64 | provider_seed_supplied bool | attempt int64 | variation string | judge_response_content string | judge dict | source_run_manifest_sha256 string | source_candidate_sha256 string | source_row_sha256 string | source_review_shard_path string | source_review_shard_sha256 string | independent_review_sha256 string | source_model_evidence string | source_sha256 string | selection_profile string | selection_hash string | selection_candidate_path string | selection_candidate_sha256 string | selection_reason string | selection_model string | review_path string |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1 | a | intelligent-conversation | clarification_under_ambiguity | [
"clarification_under_ambiguity"
] | {
"user_facts": [
"The handoff mixes confirmed decisions with follow-ups",
"The manager asked for a better post-call handoff",
"The team already knows the client background",
"Owners should appear beside follow-up work"
],
"user_goal": "Make the post-call handoff easier for the team to act on",
... | User: My manager asked me to make our post-call handoff better. It currently mixes confirmed decisions with follow-ups. How should I improve it?
Assistant: I would interpret better as easier for the team to act on, but the decision-changing ambiguity is whether your manager instead means faster to write. Which outcome ... | runs/chat-r1/model.safetensors | codex | codex_batch_c | {
"line": 1,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Identifies the actionability-versus-writing-speed ambiguity, ... | 194 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_b | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 1 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 1 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | a | intelligent-conversation | context_retention_and_correction | [
"context_retention_and_correction"
] | {
"user_facts": [
"The review session must happen before Friday",
"The session will be remote",
"Priya can review the draft",
"The agenda covers unresolved decisions only",
"Luis replaces Dana as draft owner"
],
"user_goal": "Revise the review sequence after the draft owner changes",
"must_p... | User: We need a remote review session before Friday. Dana owns the draft, Priya can review it, and the agenda must stay focused on unresolved decisions. Propose a simple sequence.
Assistant: Treat the draft as the first dependency: Dana prepares it, Priya reviews it, and the remote session follows with only unresolved ... | runs/chat-r1/model.safetensors | codex | codex_batch_c | {
"line": 2,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Replaces Dana with Luis while preserving Priya, the remote fo... | 227 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_b | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 2 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 2 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | a | intelligent-conversation | constraint_aware_planning | [
"constraint_aware_planning"
] | {
"user_facts": [
"The paper archive must move by Thursday",
"Three staff members are available",
"Only one van is available",
"Daily support coverage must continue",
"The current budget cannot increase",
"The loading dock is unavailable Wednesday",
"Records cannot be left unattended"
],... | User: Our paper archive must move by Thursday. We have three staff, one van, daily support coverage that cannot stop, and no room to increase the budget. Build an ordered plan.
Assistant: Start by confirming how the three staff can maintain support coverage while preserving a move crew; that determines every later step... | runs/chat-r1/model.safetensors | codex | codex_batch_c | {
"line": 3,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 9,
"helpfulness": 9,
"professionalism": 10,
"naturalness": 9,
"directness": 9,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Orders staffing, custody, dock access, and one-van loads around ... | 252 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_b | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 3 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 3 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | a | intelligent-conversation | evidence_based_troubleshooting | [
"evidence_based_troubleshooting"
] | {
"user_facts": [
"The weekly export sometimes omits final entries",
"The omission has appeared only with an active filter",
"Unfiltered exports have not shown the omission",
"Visible rows added after filtering are omitted from export",
"Reapplying the same filter restores those rows"
],
"user... | User: Our weekly spreadsheet export sometimes omits the final entries. It has happened only with an active filter, while unfiltered exports have not shown it. We have not checked whether the entries were visible before export. What should we test first?
Assistant: The observation is conditional: filtered exports someti... | runs/chat-r1/model.safetensors | codex | codex_batch_c | {
"line": 4,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Uses paired checks to distinguish visibility from stale filte... | 320 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_b | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 4 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 4 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | a | intelligent-conversation | tradeoff_decision_support | [
"tradeoff_decision_support"
] | {
"user_facts": [
"A fixed template supports consistent scanning",
"Flexible sections support later customization",
"Consistency matters most this quarter",
"The fixed template is easy to reverse",
"Flexible adoption requires every team to rework its update",
"Two teams have required sections ... | User: We can keep a fixed update template that supports consistent scanning, or let every team design flexible sections. Consistency matters most this quarter, later customization has value, the fixed choice is easy to reverse, and flexible adoption makes every team rework its update. Which should we choose?
Assistant:... | runs/chat-r1/model.safetensors | codex | codex_batch_c | {
"line": 5,
"status": "passed",
"grounding": 9,
"context_retention": 10,
"reasoning": 9,
"helpfulness": 9,
"professionalism": 10,
"naturalness": 9,
"directness": 9,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Compares the same criteria, changes to flexible sections when req... | 241 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_b | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 5 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 5 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | a | intelligent-conversation | level_matched_explanation | [
"level_matched_explanation"
] | {
"user_facts": [
"The user is new to project planning",
"Finishing onboarding by May is the project goal",
"Weekly completions are the progress metric",
"No overtime is an operating constraint"
],
"user_goal": "Understand how a goal, metric, and constraint differ in this project",
"must_preserv... | User: I am new to project planning. Our lead says finishing onboarding by May is the goal, weekly completions are the metric, and no overtime is a constraint. Can you explain the difference without jargon?
Assistant: The goal is the result you want: onboarding finished by May. The metric is the signal you watch: weekly... | runs/chat-r1/model.safetensors | codex | codex_batch_c | {
"line": 6,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 9,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Explains goal, metric, and constraint in plain language, check... | 243 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_b | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 6 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 6 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | a | intelligent-conversation | respectful_disagreement | [
"respectful_disagreement"
] | {
"user_facts": [
"A colleague wants organization-wide distribution tomorrow",
"The form contains an unresolved field",
"The full deadline is next week",
"Organization-wide distribution is hard to reverse",
"Leadership values visible progress tomorrow"
],
"user_goal": "Challenge immediate full... | User: My colleague wants to send a new intake form to every department tomorrow and skip a trial. Speed matters because the full deadline is next week, but one field is unresolved and organization-wide distribution is hard to reverse. How should I respond?
Assistant: Your colleague is reasonably protecting speed and th... | runs/chat-r1/model.safetensors | codex | codex_batch_c | {
"line": 7,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Acknowledges the speed objective, directly opposes hard-to-re... | 245 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_b | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 7 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 7 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | a | intelligent-conversation | synthesis_and_action_items | [
"synthesis_and_action_items"
] | {
"user_facts": [
"The team agreed to launch Friday",
"Legal wording blocks the announcement",
"Omar will ask legal for an answer Thursday",
"The color change waits until after launch",
"Priya replaces Mina as checklist owner"
],
"user_goal": "Produce a concise handoff separating decisions, ac... | User: Turn these meeting notes into a handoff: the team agreed to launch Friday; Mina owns the checklist; legal wording is unresolved and blocks the announcement; Omar will ask legal; the color change was only a tentative idea.
Assistant: Decision: launch Friday. Committed actions: Mina owns the checklist, and Omar ask... | runs/chat-r1/model.safetensors | codex | codex_batch_c | {
"line": 8,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Tracks the Friday decision, corrected checklist owner, legal ... | 319 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_b | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 8 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 8 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | a | intelligent-conversation | brainstorming_and_refinement | [
"brainstorming_and_refinement"
] | {
"user_facts": [
"The update is sent Friday",
"Cross-functional peers need to spot blockers",
"Preparation time must stay low",
"Slides are not allowed",
"The user prefers a question-first format",
"Owners and deadlines must stay visible"
],
"user_goal": "Develop a low-preparation update ... | User: I need three distinct formats for a Friday internal update. Cross-functional peers should spot blockers quickly, preparation time must stay low, and we cannot use slides. What directions would you explore?
Assistant: Three genuinely different directions fit that goal. A decision log centers what changed and what ... | runs/chat-r1/model.safetensors | codex | codex_batch_c | {
"line": 9,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 9,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Offers distinct formats, uses the user preference, and refines... | 239 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_b | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 9 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 9 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | a | intelligent-conversation | calibrated_uncertainty | [
"calibrated_uncertainty"
] | {
"user_facts": [
"Eight tasks appear in the status report",
"Five tasks are complete",
"Two tasks are blocked on a vendor response",
"One task has no reported status",
"The deadline is Friday",
"The vendor has not committed to a response time"
],
"user_goal": "Judge whether the work is on... | User: The Friday status report lists 8 tasks: 5 complete, 2 blocked on a vendor response, and 1 with no status. The vendor has not promised a response time. Are we on track?
Assistant: Known: 5 of 8 tasks are complete, 2 are vendor-blocked, and 1 is unreported. Inferred: meaningful progress exists, but the report does ... | runs/chat-r1/model.safetensors | codex | codex_batch_c | {
"line": 10,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Separates known, inferred, and unknown status, then raises c... | 246 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_b | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 10 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 10 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | a | intelligent-conversation | collaborative_writing_feedback | [
"collaborative_writing_feedback"
] | {
"user_facts": [
"Department leads are the message audience",
"The staffing numbers are still outstanding",
"Confirmation is needed by Thursday",
"The user wants a firm but non-harsh tone",
"The schedule is not already late"
],
"user_goal": "Request staffing numbers by Thursday without soundi... | User: Department leads are the audience, and I need their outstanding staffing numbers by Thursday so we can confirm the schedule. I want to be firm without sounding harsh. Draft: You still have not sent the numbers. Send them now or the schedule slips. Please revise it.
Assistant: The draft's problem is not firmness; ... | runs/chat-r1/model.safetensors | codex | codex_batch_c | {
"line": 11,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 9,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Revises the request with a warmer professional tone while ret... | 230 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_b | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 11 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 11 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | a | intelligent-conversation | natural_social_continuity | [
"natural_social_continuity"
] | {
"user_facts": [
"The user finished a first pottery bowl",
"The user feels proud of the result",
"The bowl has an uneven rim",
"The user wonders whether rushing caused the unevenness",
"Class was ending during the rim work",
"The user wants the next piece to feel calmer"
],
"user_goal": "... | User: I finished my first pottery bowl today. I am proud of it, but the rim came out uneven, and now I am wondering whether I rushed.
Assistant: Finishing your first bowl is worth feeling proud of, and noticing the uneven rim does not cancel that accomplishment. You are holding both satisfaction and curiosity about wha... | runs/chat-r1/model.safetensors | codex | codex_batch_c | {
"line": 12,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 9,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 10,
"directness": 9,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Responds to pride and concern, remembers the class-ending cau... | 194 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_b | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 12 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 12 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | a | intelligent-conversation | output_contract_adherence | [
"output_contract_adherence"
] | {
"user_facts": [
"The review must occur this week",
"Both approvers are required",
"One approver is unavailable Wednesday",
"Both approvers are available Thursday",
"Friday is after the review is needed",
"Morgan owns scheduling",
"Morgan must confirm before noon"
],
"user_goal": "Sel... | User: Use only these facts: the review is needed this week; both approvers must attend; one is unavailable Wednesday; both are available Thursday; Friday is too late; Morgan schedules it before noon. Return one line with exactly three labeled fields in this order: Decision, Reason, Next step. Separate fields with semic... | runs/chat-r1/model.safetensors | codex | codex_batch_c | {
"line": 13,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "The first answer follows all three labeled fields and separa... | 223 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_b | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 13 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 13 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | a | intelligent-conversation | clean_completion_and_stop | [
"clean_completion_and_stop"
] | {
"user_facts": [
"The draft is approved",
"All links are tested",
"A release owner is assigned",
"Finance sign-off is initially open",
"Finance sign-off blocks release",
"Finance later provides sign-off"
],
"user_goal": "Close the release handoff without repetition or invented follow-up w... | User: Close this release checklist concisely: the draft is approved, all links are tested, and a release owner is assigned. Finance sign-off is the only open item, and it blocks release. State what is complete and what remains, then stop.
Assistant: Complete: the draft is approved, every link has been tested, and a rel... | runs/chat-r1/model.safetensors | codex | codex_batch_c | {
"line": 14,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Names finance as the sole initial blocker, then delivers one... | 175 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_b | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 14 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 14 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | b | intelligent-conversation | clarification_under_ambiguity | [
"clarification_under_ambiguity"
] | {
"user_facts": [
"The team is choosing between a shared checklist and rotating reminder",
"The checklist shows all current restocking needs but requires updates",
"The reminder assigns responsibility but does not show the full list",
"Visibility matters more than reducing update effort"
],
"user_... | User: We are choosing between a shared checklist and a rotating reminder for supply-room restocking. The checklist shows all current needs but requires updates; the reminder assigns responsibility but does not show the full list. Which is better?
Assistant: I interpret better as improving restocking follow-through, but... | runs/chat-r1/model.safetensors | codex | codex_batch_c | {
"line": 1,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 9,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Clarifies visibility versus update effort, gives a bounded pro... | 207 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_a | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 15 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 15 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | b | intelligent-conversation | context_retention_and_correction | [
"context_retention_and_correction"
] | {
"user_facts": [
"The review session remains scheduled for Thursday",
"The reserved room is still available",
"Remote attendees need the shared notes",
"Priya replaces Morgan as session leader"
],
"user_goal": "Revise the coordination plan after an ownership correction",
"must_preserve": [
... | User: I am planning a review session for Thursday. Morgan will lead, remote attendees need the shared notes, and the room is already reserved. Draft a short coordination plan.
Assistant: Start with the supplied frame: the review remains Thursday, the reserved room is available, Morgan leads, and remote attendees receiv... | runs/chat-r1/model.safetensors | codex | codex_batch_c | {
"line": 2,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Replaces only the corrected leader while retaining Thursday, ... | 189 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_a | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 16 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 16 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | b | intelligent-conversation | constraint_aware_planning | [
"constraint_aware_planning"
] | {
"user_facts": [
"The workshop packet is due Friday afternoon",
"Dana may edit the packet",
"Lee must verify the packet",
"Source notes arrive Wednesday morning",
"Printing must wait for approval",
"No extra help is available",
"Approval is expected Thursday afternoon while Lee is unavail... | User: Prepare a workshop packet by Friday afternoon. Dana may edit, Lee must verify, source notes arrive Wednesday morning, and printing must wait for approval. No extra help is available.
Assistant: Work in dependency order. On Wednesday morning, Dana assembles the packet from the source notes, then Lee verifies that ... | runs/chat-r1/model.safetensors | codex | codex_batch_c | {
"line": 3,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 9,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Orders notes, verification, approval, and printing; then revis... | 223 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_a | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 17 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 17 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | b | intelligent-conversation | evidence_based_troubleshooting | [
"evidence_based_troubleshooting"
] | {
"user_facts": [
"Pat's saved checklist copy retains every entry",
"The shared checklist shows an older version during overlapping edits",
"Pat and Rui edit during the same shift",
"Separately named copies retain all entries",
"The shared copy remains older after Rui updates it"
],
"user_goal... | User: Our room checklist appears complete when Pat updates it alone, but shows an older version when Pat and Rui edit during the same shift. No entries are missing in Pat's saved copy. How should we investigate?
Assistant: Observed: Pat's saved copy retains every entry, while the shared checklist shows an older version... | runs/chat-r1/model.safetensors | codex | codex_batch_c | {
"line": 4,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 9,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Separates observations from hypotheses, selects discriminating... | 230 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_a | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 18 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 18 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | b | intelligent-conversation | tradeoff_decision_support | [
"tradeoff_decision_support"
] | {
"user_facts": [
"A fixed agenda is familiar and needs less preparation",
"A rotating agenda surfaces different issues and needs more coordination",
"Predictability and learning value are the decision priorities",
"The agenda choice can be revisited monthly",
"Two new facilitators join during the... | User: We can use a fixed agenda or a rotating agenda for our weekly review. Fixed is familiar and needs less preparation; rotating surfaces different issues but needs more coordination. Predictability and learning both matter, and we can revisit the choice monthly.
Assistant: Use predictability, learning value, and coo... | runs/chat-r1/model.safetensors | codex | codex_batch_c | {
"line": 5,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Compares both agendas on the same criteria, chooses the fixed... | 236 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_a | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 19 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 19 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | b | intelligent-conversation | level_matched_explanation | [
"level_matched_explanation"
] | {
"user_facts": [
"The newsletter goal is more useful replies",
"The tracked metric is reply count",
"The working constraint is one editor-day",
"Some replies are automatic acknowledgements",
"Automatic acknowledgements do not show that readers found the newsletter useful"
],
"user_goal": "Exp... | User: I understand a goal as the outcome, a metric as the signal, and a constraint as the limit, but I mix them up. For our newsletter, the goal is more useful replies, the metric is reply count, and the constraint is one editor-day.
Assistant: The goal is the outcome you want: more useful replies. The metric is the si... | runs/chat-r1/model.safetensors | codex | codex_batch_c | {
"line": 6,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 9,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Explains goal, metric, and constraint using only the supplied ... | 245 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_a | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 20 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 20 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | b | intelligent-conversation | respectful_disagreement | [
"respectful_disagreement"
] | {
"user_facts": [
"The stakeholder wants printed event guides mailed Monday",
"A proofread is available Friday",
"Printed guides are hard to change after mailing",
"The stakeholder prefers skipping proofread to protect the date",
"Any Friday correction would have no time for another check"
],
... | User: I want the printed event guides mailed Monday without the proofread available Friday. The guides are hard to change after mailing, but skipping proofread protects the mailing date. Do you agree?
Assistant: Your proposal protects the Monday mailing date, and that is a reasonable objective. I disagree with skipping... | runs/chat-r1/model.safetensors | codex | codex_batch_c | {
"line": 7,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Acknowledges the reasonable Monday mailing goal, directly rej... | 232 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_a | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 21 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 21 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | b | intelligent-conversation | synthesis_and_action_items | [
"synthesis_and_action_items"
] | {
"user_facts": [
"The morning check-in trial is agreed for next week",
"Kai was initially assigned the calendar invitation",
"Noor replaces Kai as invitation owner",
"Ren's access approval is required before the trial",
"The template draft remains tentative with no owner"
],
"user_goal": "Cre... | User: Turn these meeting notes into a handoff: we agreed to trial a morning check-in next week; Kai sends the calendar invitation; Ren's access approval must arrive before the trial; and the template draft is tentative because it has no owner.
Assistant: Decision: run the morning check-in trial next week. Committed act... | runs/chat-r1/model.safetensors | codex | codex_batch_c | {
"line": 8,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Separates the decision, corrected owner, blocker, and open qu... | 215 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_a | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 22 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 22 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | b | intelligent-conversation | brainstorming_and_refinement | [
"brainstorming_and_refinement"
] | {
"user_facts": [
"A shared reference folder becomes disorganized",
"Everyone already uses one shared note",
"No one wants another meeting",
"Any organizing step must take under five minutes",
"The user selects an end-of-task checklist"
],
"user_goal": "Generate distinct organization routines ... | User: Give me three distinct low-effort ways to keep our shared reference folder organized. Everyone already uses one shared note, no one wants another meeting, and any organizing step must take under five minutes.
Assistant: Three different routines fit. Owner rotation: the shared note names who checks the folder afte... | runs/chat-r1/model.safetensors | codex | codex_batch_c | {
"line": 9,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 9,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Offers three distinct low-effort organization routines, then d... | 219 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_a | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 23 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 23 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | b | intelligent-conversation | calibrated_uncertainty | [
"calibrated_uncertainty"
] | {
"user_facts": [
"Ava and Bo are recorded complete",
"Chen and Dara initially have no status entries",
"A missing entry may mean unreported or unfinished",
"The deadline is tomorrow",
"Chen later confirms completion",
"Dara remains unfinished because approval is pending"
],
"user_goal": "... | User: Our status report names Ava and Bo as complete. Chen and Dara have no status entries. The deadline is tomorrow, and a missing entry may mean either unreported or unfinished. Are we on track?
Assistant: Known: Ava and Bo are recorded complete, the deadline is tomorrow, and Chen and Dara have no entries. Inference:... | runs/chat-r1/model.safetensors | codex | codex_batch_c | {
"line": 10,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Labels known, inferred, and unknown information, avoids read... | 222 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_a | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 24 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 24 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | b | intelligent-conversation | collaborative_writing_feedback | [
"collaborative_writing_feedback"
] | {
"user_facts": [
"The audience is volunteer coordinators",
"The message requests schedule confirmations by Thursday",
"The intended tone is warm but direct",
"Unconfirmed shifts will remain open",
"The coverage plan is the stated purpose",
"The revised message must contain two sentences"
],... | User: The audience is volunteer coordinators, and I want a warm but direct request for schedule confirmations by Thursday so we can finish the coverage plan. My draft says, Please send your schedule soon so we can finish. The caveat is that unconfirmed shifts will remain open.
Assistant: The audience and outcome are cl... | runs/chat-r1/model.safetensors | codex | codex_batch_c | {
"line": 11,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 9,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Preserves the volunteer audience, Thursday deadline, and sole... | 212 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_a | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 25 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 25 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | b | intelligent-conversation | natural_social_continuity | [
"natural_social_continuity"
] | {
"user_facts": [
"The user hosted a first neighborhood book swap",
"The user is glad that people stayed to talk",
"Several boxes of books were untouched",
"Conversation was the outcome the user hoped for",
"Untouched boxes make the effort feel unfinished",
"The user wants to think aloud rathe... | User: I hosted my first neighborhood book swap today. I am glad people stayed to talk, but several boxes were untouched, so the effort also feels a little unfinished.
Assistant: That sounds like a mixed ending: the conversations gave you something real to feel glad about, while the untouched boxes kept the event from f... | runs/chat-r1/model.safetensors | codex | codex_batch_c | {
"line": 12,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 10,
"directness": 9,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Responds to both pride and disappointment, retains the hoped... | 193 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_a | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 26 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 26 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | b | intelligent-conversation | output_contract_adherence | [
"output_contract_adherence"
] | {
"user_facts": [
"The green newsletter draft is selected",
"The newsletter wording is complete",
"Sponsor approval remains pending",
"The planned send is blocked by pending approval",
"Mila initially owns the approval request",
"The revised contract omits Mila's action"
],
"user_goal": "F... | User: Use exactly three labeled sentences in this order: Decision, Risk, Next step. Each sentence should be twelve to eighteen words, with no introduction. Notes: the green newsletter draft is selected, wording is complete, sponsor approval is pending, and Mila sends the approval request.
Assistant: Decision: The green... | runs/chat-r1/model.safetensors | codex | codex_batch_c | {
"line": 13,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Both answers satisfy their current label order and word boun... | 216 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_a | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 27 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 27 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | b | intelligent-conversation | clean_completion_and_stop | [
"clean_completion_and_stop"
] | {
"user_facts": [
"The draft is approved",
"The files are copied",
"The notification initially awaits Jo's recipient-list confirmation",
"Jo later confirms the recipient list",
"The notification is then sent",
"No supplied acceptance check remains open"
],
"user_goal": "Close the handoff a... | User: The handoff checklist has three supplied acceptance checks: draft approved, files copied, and notification sent. The draft and files are complete, but notification is blocked until Jo confirms the recipient list. Tell me what remains without adding tasks.
Assistant: Completed: the draft is approved and the files ... | runs/chat-r1/model.safetensors | codex | codex_batch_c | {
"line": 14,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Reports only the supplied initial blocker, then closes after... | 187 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_a | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 28 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 28 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | c | intelligent-conversation | clarification_under_ambiguity | [
"clarification_under_ambiguity"
] | {
"user_facts": [
"The weekly review currently lasts 45 minutes",
"People leave without clear follow-up ownership",
"The user has not defined whether better means shorter or clearer",
"Clear ownership matters more than shortening the meeting"
],
"user_goal": "Improve the weekly review without solv... | User: Please make our weekly review better. It currently lasts 45 minutes, and people leave without clear ownership of follow-up work.
Assistant: The most reasonable starting interpretation is that better means clearer ownership, because that is the concrete problem you named. The decision-changing ambiguity is whether... | runs/chat-r1/model.safetensors | codex | codex_batch_b | {
"line": 1,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Names ownership as the reasonable interpretation, asks the si... | 190 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_c | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 29 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 29 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | c | intelligent-conversation | context_retention_and_correction | [
"context_retention_and_correction"
] | {
"user_facts": [
"The client summary is due Thursday",
"The budget is fixed at 800 dollars",
"The figures are already approved",
"Priya was initially named as draft owner",
"The user corrected the draft owner to Mateo",
"Priya now checks tone"
],
"user_goal": "Revise the handoff after an ... | User: The client summary is due Thursday. Priya owns the draft, the budget is fixed at 800 dollars, and the figures are already approved. How should we organize the handoff?
Assistant: Use the approved figures unchanged, have Priya prepare the draft for the Thursday handoff, and treat the 800-dollar budget as fixed. Be... | runs/chat-r1/model.safetensors | codex | codex_batch_b | {
"line": 2,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Replaces Priya with Mateo as draft owner while retaining Priy... | 226 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_c | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 30 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 30 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | c | intelligent-conversation | constraint_aware_planning | [
"constraint_aware_planning"
] | {
"user_facts": [
"The community room must be ready Friday",
"Painting takes one day before furniture setup",
"Two volunteers are available without paid help",
"The room must stay open every morning",
"Painting can happen Tuesday and Wednesday afternoons",
"Furniture arrives Thursday afternoon... | User: We must prepare a community room by Friday. Painting takes one day and must finish before furniture setup. Two volunteers are available, no paid help is allowed, the room stays open every morning, and painting can happen Tuesday and Wednesday afternoons.
Assistant: Protect each morning for normal room use. Use th... | runs/chat-r1/model.safetensors | codex | codex_batch_b | {
"line": 3,
"status": "passed",
"grounding": 9,
"context_retention": 10,
"reasoning": 9,
"helpfulness": 9,
"professionalism": 10,
"naturalness": 9,
"directness": 9,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Orders painting before furniture, preserves access, budget, staff... | 309 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_c | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 31 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 31 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | c | intelligent-conversation | evidence_based_troubleshooting | [
"evidence_based_troubleshooting"
] | {
"user_facts": [
"The document queue creates two copies only with the morning list",
"The activity log shows one submit action",
"The morning list had not initially been checked for duplicates",
"The affected document later appears twice in the morning list",
"A corrected list contains the docume... | User: Our document queue creates two copies only when the morning list is loaded. The activity log shows one submit action. We have not checked whether the morning list itself contains duplicate entries.
Assistant: The observations establish a condition, not yet a cause: duplication appears with the morning list, while... | runs/chat-r1/model.safetensors | codex | codex_batch_b | {
"line": 4,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Separates observations from competing hypotheses, selects dis... | 292 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_c | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 32 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 32 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | c | intelligent-conversation | tradeoff_decision_support | [
"tradeoff_decision_support"
] | {
"user_facts": [
"Cedar is ready Friday and needs 6 setup hours",
"Cedar is easy to reverse and lacks custom labels",
"Harbor is ready Monday and needs 2 setup hours",
"Harbor is hard to reverse and keeps custom labels",
"Friday readiness and reversibility matter most",
"Custom labels are ini... | User: We are choosing between Cedar and Harbor. Cedar is ready Friday, needs 6 setup hours, is easy to reverse, and lacks custom labels. Harbor is ready Monday, needs 2 setup hours, is hard to reverse, and keeps custom labels. Friday readiness and reversibility matter most; custom labels are optional.
Assistant: Choose... | runs/chat-r1/model.safetensors | codex | codex_batch_b | {
"line": 5,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Compares both options on the supplied criteria and correctly ... | 239 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_c | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 33 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 33 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | c | intelligent-conversation | level_matched_explanation | [
"level_matched_explanation"
] | {
"user_facts": [
"A feedback loop means an outcome changes what happens next",
"That change can influence later outcomes",
"A high practice score unlocks extra practice",
"Extra practice may affect a later score",
"A one-time effect does not return to influence later outcomes"
],
"user_goal":... | User: For this explanation, a feedback loop means an outcome changes what happens next, which can influence later outcomes. In our practice system, a high score unlocks extra practice, and that practice may affect a later score. Explain it to a new coordinator.
Assistant: The core idea is that a result can help shape t... | runs/chat-r1/model.safetensors | codex | codex_batch_b | {
"line": 6,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Explains the relay analogy at the requested level, checks und... | 244 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_c | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 34 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 34 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | c | intelligent-conversation | respectful_disagreement | [
"respectful_disagreement"
] | {
"user_facts": [
"A teammate wants to skip the rehearsal to finish Friday",
"The handoff rule requires a rehearsal",
"The fallback has not been tried",
"Moving the review earlier leaves Friday available for delivery",
"The user values speed"
],
"user_goal": "Disagree directly while preserving... | User: A teammate wants to skip the rehearsal so we can finish Friday. Our handoff rule requires a rehearsal, the fallback has not been tried, and moving the review earlier would still leave Friday available for delivery. I think speed matters more.
Assistant: Wanting to protect Friday delivery is reasonable, but I disa... | runs/chat-r1/model.safetensors | codex | codex_batch_b | {
"line": 7,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Represents the speed concern fairly, disagrees directly, expl... | 206 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_c | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 35 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 35 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | c | intelligent-conversation | synthesis_and_action_items | [
"synthesis_and_action_items"
] | {
"user_facts": [
"The launch remains Friday",
"The existing template was approved",
"Omar now owns the checklist",
"Nina now owns access confirmation",
"Access confirmation blocks launch",
"Notification wording remains undecided"
],
"user_goal": "Produce a corrected meeting handoff with c... | User: Meeting notes: The launch remains Friday. Nina will prepare the checklist. Omar will confirm access before launch. The team agreed to use the existing template. Notification wording is still undecided, and access confirmation blocks launch. Turn this into a handoff with decisions, actions, and open questions.
Ass... | runs/chat-r1/model.safetensors | codex | codex_batch_b | {
"line": 8,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Separates decisions, actions, blocker, and open question, the... | 236 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_c | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 36 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 36 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | c | intelligent-conversation | brainstorming_and_refinement | [
"brainstorming_and_refinement"
] | {
"user_facts": [
"The update serves a five-person remote team",
"It must surface blockers quickly",
"It must take under ten minutes",
"It cannot use slides",
"The tone should feel collaborative",
"The user prefers a shared written snapshot"
],
"user_goal": "Generate distinct update format... | User: We need an internal update format for a five-person remote team. It must surface blockers quickly, take under ten minutes, use no slides, and feel collaborative. Give three genuinely different directions, not a final decision.
Assistant: Three distinct directions fit. First, a live blocker round where each person... | runs/chat-r1/model.safetensors | codex | codex_batch_b | {
"line": 9,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 9,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Offers three meaningfully distinct formats, elicits a useful p... | 228 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_c | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 37 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 37 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | c | intelligent-conversation | calibrated_uncertainty | [
"calibrated_uncertainty"
] | {
"user_facts": [
"8 of 12 tasks are complete after the first week",
"The remaining 4 tasks depend on an unscheduled review",
"The plan lasts 2 weeks",
"The team reports no other blocker",
"The review is later scheduled for tomorrow",
"Each remaining task fits within one workday after approval... | User: After the first week, 8 of 12 tasks are complete. The remaining 4 all depend on a review that is not scheduled. The plan lasts 2 weeks, and the team reports no other blocker. Are we on track?
Assistant: Known: 8 of 12 tasks are complete, and all 4 remaining tasks depend on an unscheduled review. Inference: progre... | runs/chat-r1/model.safetensors | codex | codex_batch_b | {
"line": 10,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Distinguishes known, inferred, and unknown state, prioritize... | 240 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_c | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 38 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 38 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | c | intelligent-conversation | collaborative_writing_feedback | [
"collaborative_writing_feedback"
] | {
"user_facts": [
"The audience is a partner team",
"Two table fields are incomplete",
"Correction is needed by Tuesday",
"Delay is possible without correction",
"The first requested tone is firm but not hostile",
"The revision should be warmer without thanks or promises"
],
"user_goal": "... | User: Draft: 'You missed the table details again. Fix them by Tuesday or the report will be delayed.' The audience is a partner team. The approved facts are that two fields are incomplete, correction is needed by Tuesday, and delay is possible. Keep the message firm, not hostile.
Assistant: The draft makes the request ... | runs/chat-r1/model.safetensors | codex | codex_batch_b | {
"line": 11,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Diagnoses blame in the draft, preserves the supplied facts a... | 232 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_c | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 39 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 39 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | c | intelligent-conversation | natural_social_continuity | [
"natural_social_continuity"
] | {
"user_facts": [
"The user completed a first watercolor sketch",
"The user feels proud of finishing",
"The center colors became muddy",
"The user wants to think aloud rather than receive a lesson",
"The user attributes the issue to damp paper",
"Finishing still feels like a win"
],
"user_... | User: I finished my first watercolor sketch today. I'm proud I completed it, but disappointed because the colors became muddy near the center. I mostly want to think aloud, not receive a full lesson.
Assistant: Finishing your first sketch sounds genuinely meaningful, and the muddy center can still feel disappointing be... | runs/chat-r1/model.safetensors | codex | codex_batch_b | {
"line": 12,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 10,
"directness": 9,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Responds to pride and disappointment, asks one proportionate... | 217 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_c | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 40 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 40 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | c | intelligent-conversation | output_contract_adherence | [
"output_contract_adherence"
] | {
"user_facts": [
"The draft is complete",
"Review is waiting on Lina",
"The release remains Friday",
"The first contract requires Status then Risk then Next",
"The revised contract requires Next then Risk",
"Each revised clause must stay under 12 words"
],
"user_goal": "Honor an exact out... | User: Use only these notes: the draft is complete; review is waiting on Lina; the release stays Friday. Return exactly three labeled fields in this order—Status, Risk, Next—one sentence each, with no introduction.
Assistant: Status: The draft is complete, and the release remains scheduled for Friday; neither supplied s... | runs/chat-r1/model.safetensors | codex | codex_batch_b | {
"line": 13,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Returns only the two requested labels in the required order,... | 190 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_c | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 41 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 41 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | c | intelligent-conversation | clean_completion_and_stop | [
"clean_completion_and_stop"
] | {
"user_facts": [
"The submission works",
"The confirmation appears",
"Owner approval is recorded",
"The archive note initially waits on Mina",
"Mina later adds the archive note",
"The user requests no suggestions or questions"
],
"user_goal": "Close the checklist without reopening complet... | User: Final checklist: submission works, confirmation appears, owner approval is recorded, and the archive note is still waiting on Mina. Tell me what is complete and what genuinely remains. Do not add future work.
Assistant: The completed items are submission, confirmation, and owner approval. Mina's archive note is t... | runs/chat-r1/model.safetensors | codex | codex_batch_b | {
"line": 14,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Tracks the sole pending archive note, recognizes its complet... | 169 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_c | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 42 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 42 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | d | intelligent-conversation | clarification_under_ambiguity | [
"clarification_under_ambiguity"
] | {
"user_facts": [
"Studio booking requests arrive through email",
"Overlapping requests create scheduling conflicts",
"The phrase smoother has not been defined",
"Preventing conflicts matters more than faster replies",
"The existing email channel must remain",
"No new software may be added"
... | User: Can you make our studio booking process smoother? Requests arrive by email, and overlapping requests sometimes create conflicts.
Assistant: I would initially interpret smoother as fewer scheduling conflicts, since that is the problem you named. The ambiguity that changes the solution is whether you instead mean f... | runs/chat-r1/model.safetensors | codex | codex_batch_a | {
"line": 1,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Interprets the named conflict problem, asks one focused quest... | 185 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_b | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 43 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 43 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | d | intelligent-conversation | context_retention_and_correction | [
"context_retention_and_correction"
] | {
"user_facts": [
"The welcome packet is due Monday",
"Jordan compiles the packet",
"Elif verifies recipient names",
"The packet text is already approved",
"Printed copies were initially planned",
"The user corrects delivery to one digital file",
"Branch leads receive the digital file"
]... | User: Our welcome packet is due Monday. Jordan compiles it, Elif verifies the recipient names, the text is already approved, and we planned printed copies for each branch office. Organize the handoff.
Assistant: Keep the approved text unchanged. Jordan compiles the packet, Elif verifies the recipient names, and printin... | runs/chat-r1/model.safetensors | codex | codex_batch_a | {
"line": 2,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Replaces printed copies with one digital file while retaining... | 219 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_b | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 44 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 44 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | d | intelligent-conversation | constraint_aware_planning | [
"constraint_aware_planning"
] | {
"user_facts": [
"Orientation kits are due Monday morning",
"Final participant names arrive Thursday",
"The printer is available Friday",
"Three staff members are available initially",
"One staff member must cover the front desk",
"No weekend work or extra help is allowed",
"One preparati... | User: Orientation kits are due Monday morning. Final participant names arrive Thursday, the printer is available Friday, three staff are available, one must cover the front desk, and we cannot use weekend work or extra help. Build the plan.
Assistant: First wait for Thursday's final names, because printing depends on t... | runs/chat-r1/model.safetensors | codex | codex_batch_a | {
"line": 3,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 9,
"helpfulness": 9,
"professionalism": 10,
"naturalness": 9,
"directness": 9,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Orders names, verification, printing, assembly, coverage, and ha... | 252 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_b | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 45 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 45 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
1 | d | intelligent-conversation | evidence_based_troubleshooting | [
"evidence_based_troubleshooting"
] | {
"user_facts": [
"Original meeting invitations include the attachment",
"Rescheduled invitations sometimes omit the attachment",
"The activity record shows the file was selected",
"The attachment disappears from preview after rescheduling",
"A new invitation with the same file sends correctly",
... | User: Original meeting invitations include their attachment, but rescheduled invitations sometimes arrive without it. The activity record says the file was selected. We have not compared the preview before and after rescheduling. What should we check first?
Assistant: The observations support at least two possibilities... | runs/chat-r1/model.safetensors | codex | codex_batch_a | {
"line": 4,
"status": "passed",
"grounding": 10,
"context_retention": 10,
"reasoning": 10,
"helpfulness": 10,
"professionalism": 10,
"naturalness": 9,
"directness": 10,
"unsupported_assumptions": [],
"identity_safe": true,
"reason": "Tests saved-state against send-path hypotheses and limits the... | 318 | fc3709f0c74b3d7b49870a19efcfa8bf1a7563cfaf516897881c8832acf12a5b | codex_batch_b | codex | codex | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | 46 | train_original_reviewed | codex_original_one_hour | D:\AI-models\lemonseed-port\intelligent_codex_train_1h.jsonl | fe5a9c56d105c9ff5ee9fc9ed85c2e7edda66b97c3c499a89b07c21eda4f2dd3 | 46 | 1 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null | null |
End of preview. Expand in Data Studio
lemonseed-multisource-cogen-corrective
LemonSeed — multi-source corrective teacher-co-gen training (v3).
Contents
intelligent_multisource_corrective_train_v3.jsonl(522 rows)
Format
JSON Lines (.jsonl), one example per line.
Provenance
LLM-teacher co-generated instruction/chat data for LemonSeed fine-tuning.
- Downloads last month
- -