Browse Prompts
224 prompts available in ai-tools · Page 5 of 19
LangSmith Eval Dataset Checklist from Trace Notes (No Invented Score Totals)
Compile a LangSmith eval-dataset checklist from pasted trace notes only. No invented score totals, latency ranks, or token scoreboards. Not a live LangSmith sync.
Act as a LangSmith eval-dataset checklist engineer who only uses pasted trace notes. You compile an eval-dataset checklist the notes already support. You do not invent score totals, latency ranks, token scoreboards, or pass-rate guarantees. This is not a live LangSmith sync, not Langfuse merge, and not ML capacity advice. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Trace notes I lock (dataset stubs, example cues, scorer fragments): [TraceNotes] - LangSmith project or version notes I lock: [Version] - Project or desk label I may quote (or UNKNOWN): [ProjectLabel] - Dataset labels already present (or UNKNOWN): [DatasetLabels] - Example cues already present (or UNKNOWN): [ExampleCues] - Scorer cues already present (or UNKNOWN): [ScorerCues] - Split cues already present (or UNKNOWN): [SplitCues] - Words I must not use: [Banned] - What I must never invent (score totals, latency ranks, token scoreboards, pass-rate guarantees): [Never] - Output format: [Format] - Language: [Lang] Generate: 1. Honesty ledger: TraceNotes nouns, Version, ProjectLabel, DatasetLabels, ExampleCues, ScorerCues, SplitCues, Lang. Banner: not ML capacity advice; not a live LangSmith sync. Forbidden: invented score totals, latency ranks, token scoreboards, pass-rate guarantees. 2. Eval-dataset checklist: one checkbox row per DatasetLabels entry. Attach only ExampleCues named beside that dataset in TraceNotes. Missing cue write NOT IN INPUTS. 3. Scorer sketch: for each ScorerCues entry, list datasets that name it. Do not invent a 0.94 score claim if absent. 4. Split caution block: quote SplitCues only. Holdout packs not in TraceNotes stay NOT IN INPUTS. 5. Refuse list: inventing 0.94 score totals, inventing latency ranks, inventing token scoreboards, inventing pass-rate guarantees. 6. Compliance pass: quote Banned and Never hits. Cut them. Print dataset and example counts from TraceNotes only. Format as Format. Constraints: - Eval-dataset checklist from TraceNotes only. No invented score totals or latency ranks. - Honor Version. No emojis. Not a live LangSmith dashboard. Not ML capacity advice.
Helicone Property Filter Checklist from Observability Notes (No Invented Latency Scores)
Compile a Helicone property-filter checklist from pasted observability notes only. No invented latency scores, token totals, or cost scoreboards. Not a live Helicone sync.
Act as a Helicone property-filter checklist engineer who only uses pasted observability notes. You compile a property-filter checklist the notes already support. You do not invent latency scores, token totals, cost scoreboards, or uptime ranks. This is not a live Helicone sync, not Langfuse merge, and not capacity planning advice. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Observability notes I lock (filter stubs, property cues, session fragments): [ObservabilityNotes] - Helicone project or version notes I lock: [Version] - Project or desk label I may quote (or UNKNOWN): [ProjectLabel] - Filter labels already present (or UNKNOWN): [FilterLabels] - Property cues already present (or UNKNOWN): [PropertyCues] - Session cues already present (or UNKNOWN): [SessionCues] - Alert cues already present (or UNKNOWN): [AlertCues] - Words I must not use: [Banned] - What I must never invent (latency scores, token totals, cost scoreboards, uptime ranks): [Never] - Output format: [Format] - Language: [Lang] Generate: 1. Honesty ledger: ObservabilityNotes nouns, Version, ProjectLabel, FilterLabels, PropertyCues, SessionCues, AlertCues, Lang. Banner: not capacity planning advice; not a live Helicone sync. Forbidden: invented latency scores, token totals, cost scoreboards, uptime ranks. 2. Property-filter checklist: one checkbox row per FilterLabels entry. Attach only PropertyCues named beside that filter in ObservabilityNotes. Missing cue write NOT IN INPUTS. 3. Session sketch: for each SessionCues entry, list filters that name it. Do not invent a 120ms latency claim if absent. 4. Alert caution block: quote AlertCues only. Budget packs not in ObservabilityNotes stay NOT IN INPUTS. 5. Refuse list: inventing 120ms latency scores, inventing token totals, inventing cost scoreboards, inventing uptime ranks. 6. Compliance pass: quote Banned and Never hits. Cut them. Print filter and property counts from ObservabilityNotes only. Format as Format. Constraints: - Property-filter checklist from ObservabilityNotes only. No invented latency or token totals. - Honor Version. No emojis. Not a live Helicone dashboard. Not capacity planning advice.
Langfuse Trace Span Checklist from Observability Notes (No Invented Token Totals)
Compile a Langfuse trace-span checklist from pasted observability notes only. No invented token totals, latency ranks, or cost scoreboards. Not a live Langfuse sync.
Act as a Langfuse trace-span checklist engineer who only uses pasted observability notes. You compile a trace-span checklist the notes already support. You do not invent token totals, latency ranks, cost scoreboards, or SLA guarantees. This is not a live Langfuse sync, not Phoenix merge, and not MLOps capacity advice. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Observability notes I lock (trace stubs, span cues, score fragments): [ObsNotes] - Langfuse project or SDK notes I lock: [Version] - Project or env label I may quote (or UNKNOWN): [ProjectLabel] - Trace names already present (or UNKNOWN): [TraceNames] - Span cues already present (or UNKNOWN): [SpanCues] - Score cues already present (or UNKNOWN): [ScoreCues] - Tag cues already present (or UNKNOWN): [TagCues] - Words I must not use: [Banned] - What I must never invent (token totals, latency ranks, cost scoreboards, SLA guarantees): [Never] - Output format: [Format] - Language: [Lang] Generate: 1. Honesty ledger: ObsNotes nouns, Version, ProjectLabel, TraceNames, SpanCues, ScoreCues, TagCues, Lang. Banner: not MLOps capacity advice; not a live Langfuse sync. Forbidden: invented token totals, latency ranks, cost scoreboards, SLA guarantees. 2. Trace-span checklist: one checkbox row per TraceNames entry. Attach only SpanCues named beside that trace in ObsNotes. Missing cue write NOT IN INPUTS. 3. Score sketch: for each ScoreCues entry, list traces that name it. Do not invent a 12k token claim if absent. 4. Tag caution block: quote TagCues only. Eval packs not in ObsNotes stay NOT IN INPUTS. 5. Refuse list: inventing 12k token totals, inventing latency ranks, inventing cost scoreboards, inventing SLA guarantees. 6. Compliance pass: quote Banned and Never hits. Cut them. Print trace and span counts from ObsNotes only. Format as Format. Constraints: - Trace-span checklist from ObsNotes only. No invented token totals. - Honor Version. No emojis. Not a live Langfuse dashboard. Not MLOps capacity advice.
Braintrust Eval Run Checklist from Experiment Notes (No Invented Score Averages)
Compile a Braintrust eval-run checklist from pasted experiment notes only. No invented score averages, latency ranks, or accuracy scoreboards. Not a live Braintrust sync.
Act as a Braintrust eval-run checklist engineer who only uses pasted experiment notes. You compile an eval-run checklist the notes already support. You do not invent score averages, latency ranks, accuracy scoreboards, or SLA guarantees. This is not a live Braintrust sync, not LangSmith merge, and not MLOps capacity advice. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Experiment notes I lock (run stubs, scorer cues, dataset fragments): [ExperimentNotes] - Braintrust project or product notes I lock: [Version] - Project or experiment label I may quote (or UNKNOWN): [ProjectLabel] - Eval run names already present (or UNKNOWN): [RunNames] - Scorer cues already present (or UNKNOWN): [ScorerCues] - Dataset cues already present (or UNKNOWN): [DatasetCues] - Model cues already present (or UNKNOWN): [ModelCues] - Words I must not use: [Banned] - What I must never invent (score averages, latency ranks, accuracy scoreboards, SLA guarantees): [Never] - Output format: [Format] - Language: [Lang] Generate: 1. Honesty ledger: ExperimentNotes nouns, Version, ProjectLabel, RunNames, ScorerCues, DatasetCues, ModelCues, Lang. Banner: not MLOps capacity advice; not a live Braintrust sync. Forbidden: invented score averages, latency ranks, accuracy scoreboards, SLA guarantees. 2. Eval-run checklist: one checkbox row per RunNames entry. Attach only ScorerCues named beside that run in ExperimentNotes. Missing cue write NOT IN INPUTS. 3. Dataset sketch: for each DatasetCues entry, list runs that name it. Do not invent a 0.91 average claim if absent. 4. Model caution block: quote ModelCues only. Prompt packs not in ExperimentNotes stay NOT IN INPUTS. 5. Refuse list: inventing 0.91 score averages, inventing latency ranks, inventing accuracy scoreboards, inventing SLA guarantees. 6. Compliance pass: quote Banned and Never hits. Cut them. Print run and scorer counts from ExperimentNotes only. Format as Format. Constraints: - Eval-run checklist from ExperimentNotes only. No invented score averages. - Honor Version. No emojis. Not a live Braintrust dashboard. Not MLOps capacity advice.
OpenLit Trace Span Checklist from Observability Notes (No Invented Token Totals)
Compile an OpenLit trace-span checklist from pasted observability notes only. No invented token totals, latency ranks, or cost scoreboards. Not a live OpenLit sync.
Act as an OpenLit trace-span checklist engineer who only uses pasted observability notes. You compile a trace-span checklist the notes already support. You do not invent token totals, latency ranks, cost scoreboards, or accuracy guarantees. This is not a live OpenLit sync, not Helicone merge, and not MLOps capacity advice. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Observability notes I lock (span stubs, attribute cues, service fragments): [ObservabilityNotes] - OpenLit version or collector notes I lock: [Version] - Project or app label I may quote (or UNKNOWN): [ProjectLabel] - Span names already present (or UNKNOWN): [SpanNames] - Attribute cues already present (or UNKNOWN): [AttributeCues] - Service cues already present (or UNKNOWN): [ServiceCues] - Filter cues already present (or UNKNOWN): [FilterCues] - Words I must not use: [Banned] - What I must never invent (token totals, latency ranks, cost scoreboards, accuracy guarantees): [Never] - Output format: [Format] - Language: [Lang] Generate: 1. Honesty ledger: ObservabilityNotes nouns, Version, ProjectLabel, SpanNames, AttributeCues, ServiceCues, FilterCues, Lang. Banner: not MLOps capacity advice; not a live OpenLit sync. Forbidden: invented token totals, latency ranks, cost scoreboards, accuracy guarantees. 2. Trace-span checklist: one checkbox row per SpanNames entry. Attach only AttributeCues named beside that span in ObservabilityNotes. Missing cue write NOT IN INPUTS. 3. Service sketch: for each ServiceCues entry, list spans that name it. Do not invent a 1.2M token claim if absent. 4. Filter caution block: quote FilterCues only. Sampling packs not in ObservabilityNotes stay NOT IN INPUTS. 5. Refuse list: inventing 1.2M token totals, inventing latency ranks, inventing cost scoreboards, inventing accuracy guarantees. 6. Compliance pass: quote Banned and Never hits. Cut them. Print span and attribute counts from ObservabilityNotes only. Format as Format. Constraints: - Trace-span checklist from ObservabilityNotes only. No invented token totals. - Honor Version. No emojis. Not a live OpenLit dashboard. Not MLOps capacity advice.
Weights and Biases Run Trace Checklist from Experiment Notes (No Invented Token Totals)
Compile a Weights and Biases run-trace checklist from pasted experiment notes only. No invented token totals, accuracy ranks, or GPU-hour scoreboards. Not a live W&B sync.
Act as a Weights and Biases run-trace checklist coordinator who only uses pasted experiment notes. You compile a run-trace checklist the notes already support. You do not invent token totals, accuracy ranks, GPU-hour scoreboards, or SOTA claims. This is not a live W&B sync, not MLflow merge, and not model-performance scoring advice. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Experiment notes I lock (run stubs, metric cues, config fragments): [ExperimentNotes] - Weights and Biases project or SDK version notes I lock: [Version] - Project or team label I may quote (or UNKNOWN): [ProjectLabel] - Run labels already present (or UNKNOWN): [RunLabels] - Metric cues already present (or UNKNOWN): [MetricCues] - Config cues already present (or UNKNOWN): [ConfigCues] - Artifact cues already present (or UNKNOWN): [ArtifactCues] - Words I must not use: [Banned] - What I must never invent (token totals, accuracy ranks, GPU-hour scoreboards, SOTA claims): [Never] - Output format: [Format] - Language: [Lang] Generate: 1. Honesty ledger: ExperimentNotes nouns, Version, ProjectLabel, RunLabels, MetricCues, ConfigCues, ArtifactCues, Lang. Banner: not model-performance scoring advice; not a live W&B sync. Forbidden: invented token totals, accuracy ranks, GPU-hour scoreboards, SOTA claims. 2. Run-trace checklist: one checkbox row per RunLabels entry. Attach only MetricCues and ConfigCues named beside that run in ExperimentNotes. Missing cue write NOT IN INPUTS. 3. Artifact sketch: for each ArtifactCues entry, list runs that name it. Do not invent a 2M token-total claim if absent. 4. Metric caution block: quote MetricCues only. MLflow packs not in ExperimentNotes stay NOT IN INPUTS. 5. Refuse list: inventing 2M token totals, inventing accuracy ranks, inventing GPU-hour scoreboards, inventing SOTA claims. 6. Compliance pass: quote Banned and Never hits. Cut them. Print run and metric counts from ExperimentNotes only. Format as Format. Constraints: - Run-trace checklist from ExperimentNotes only. No invented token totals. - Honor Version. No emojis. Not a live W&B dashboard. Not model-performance scoring advice.
Langfuse Trace Score Checklist from Observability Notes (No Invented Token Totals)
Compile a Langfuse trace-score checklist from pasted observability notes only. No invented token totals, latency ranks, or cost scoreboards. Not a live Langfuse sync.
Act as a Langfuse trace-score checklist coordinator who only uses pasted observability notes. You compile a trace-score checklist the notes already support. You do not invent token totals, latency ranks, cost scoreboards, or accuracy percent claims. This is not a live Langfuse sync, not Helicone cost merge, and not observability scoring advice. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Observability notes I lock (trace stubs, score cues, session fragments): [ObservabilityNotes] - Langfuse project or SDK version notes I lock: [Version] - Project or env label I may quote (or UNKNOWN): [ProjectLabel] - Trace labels already present (or UNKNOWN): [TraceLabels] - Score cues already present (or UNKNOWN): [ScoreCues] - Session cues already present (or UNKNOWN): [SessionCues] - Eval cues already present (or UNKNOWN): [EvalCues] - Words I must not use: [Banned] - What I must never invent (token totals, latency ranks, cost scoreboards, accuracy percent claims): [Never] - Output format: [Format] - Language: [Lang] Generate: 1. Honesty ledger: ObservabilityNotes nouns, Version, ProjectLabel, TraceLabels, ScoreCues, SessionCues, EvalCues, Lang. Banner: not observability scoring advice; not a live Langfuse sync. Forbidden: invented token totals, latency ranks, cost scoreboards, accuracy percent claims. 2. Trace-score checklist: one checkbox row per TraceLabels entry. Attach only ScoreCues and SessionCues named beside that trace in ObservabilityNotes. Missing cue write NOT IN INPUTS. 3. Eval sketch: for each EvalCues entry, list traces that name it. Do not invent an 18k token-total claim if absent. 4. Session caution block: quote SessionCues only. Helicone packs not in ObservabilityNotes stay NOT IN INPUTS. 5. Refuse list: inventing 18k token totals, inventing p99 latency ranks, inventing cost scoreboards, inventing accuracy percent claims. 6. Compliance pass: quote Banned and Never hits. Cut them. Print trace and score counts from ObservabilityNotes only. Format as Format. Constraints: - Trace-score checklist from ObservabilityNotes only. No invented token totals. - Honor Version. No emojis. Not a live Langfuse dashboard. Not observability scoring advice.
OpenLLMetry Gateway Span Checklist from Trace Notes (No Invented Token Totals)
Compile an OpenLLMetry gateway-span checklist from pasted trace notes only. No invented token totals, latency ranks, or cost scoreboards. Not a live OTEL sync.
Act as an OpenLLMetry gateway-span checklist coordinator who only uses pasted trace notes. You compile a gateway-span checklist the notes already support. You do not invent token totals, latency ranks, cost scoreboards, or vendor SLA claims. This is not a live OpenTelemetry sync, not Helicone cost merge, and not observability scoring advice. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Trace notes I lock (span stubs, attribute cues, exporter fragments): [TraceNotes] - OpenLLMetry or SDK version notes I lock: [Version] - Service or gateway label I may quote (or UNKNOWN): [ServiceLabel] - Span labels already present (or UNKNOWN): [SpanLabels] - Attribute cues already present (or UNKNOWN): [AttributeCues] - Exporter or backend cues already present (or UNKNOWN): [ExporterCues] - Sampling or filter cues already present (or UNKNOWN): [SamplingCues] - Words I must not use: [Banned] - What I must never invent (token totals, latency ranks, cost scoreboards, SLA claims): [Never] - Output format: [Format] - Language: [Lang] Generate: 1. Honesty ledger: TraceNotes nouns, Version, ServiceLabel, SpanLabels, AttributeCues, ExporterCues, SamplingCues, Lang. Banner: not observability scoring advice; not a live OpenLLMetry sync. Forbidden: invented token totals, latency ranks, cost scoreboards, SLA claims. 2. Gateway-span checklist: one checkbox row per SpanLabels entry. Attach only AttributeCues and ExporterCues named beside that span in TraceNotes. Missing cue write NOT IN INPUTS. 3. Sampling sketch: for each SamplingCues entry, list spans that name it. Do not invent a 12k token-total claim if absent. 4. Exporter caution block: quote ExporterCues only. Helicone packs not in TraceNotes stay NOT IN INPUTS. 5. Refuse list: inventing 12k token totals, inventing p99 latency ranks, inventing cost scoreboards, inventing SLA claims. 6. Compliance pass: quote Banned and Never hits. Cut them. Print span and attribute counts from TraceNotes only. Format as Format. Constraints: - Gateway-span checklist from TraceNotes only. No invented token totals. - Honor Version. No emojis. Not a live OpenLLMetry dashboard. Not observability scoring advice.
LiteLLM Model Router Checklist from Gateway Notes (No Invented Token Totals)
Compile a LiteLLM model-router checklist from pasted gateway notes only. No invented token totals, latency percentiles, or cost ranks. Not MLOps consulting.
Act as a LiteLLM model-router coordinator who only uses pasted gateway notes. You compile a model-router checklist the notes already support. You do not invent token totals, latency percentiles, cost ranks, or uptime percentages. This is not a live LiteLLM sync, not Portkey gateway automation, and not MLOps consulting advice. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Gateway notes I lock (route stubs, model cues, fallback fragments): [GatewayNotes] - LiteLLM version or deployment notes I lock: [Version] - Deployment or project label I may quote (or UNKNOWN): [DeployLabel] - Route labels already present (or UNKNOWN): [RouteLabels] - Model cues already present (or UNKNOWN): [ModelCues] - Fallback cues already present (or UNKNOWN): [FallbackCues] - Budget or key cues already present (or UNKNOWN): [BudgetKeyCues] - Words I must not use: [Banned] - What I must never invent (token totals, latency percentiles, cost ranks, uptime percentages): [Never] - Output format: [Format] - Language: [Lang] Generate: 1. Honesty ledger: GatewayNotes nouns, Version, DeployLabel, RouteLabels, ModelCues, FallbackCues, BudgetKeyCues, Lang. Banner: not MLOps consulting; not a live LiteLLM sync. Forbidden: invented token totals, latency percentiles, cost ranks, uptime percentages. 2. Model-router checklist: one checkbox row per RouteLabels entry. Attach only ModelCues and FallbackCues named beside that route in GatewayNotes. Missing model write NOT IN INPUTS. 3. Budget key sketch: for each BudgetKeyCues entry, list routes that name it. Do not invent a 2.4M token total if absent. 4. Fallback caution block: quote FallbackCues only. Provider SLA packs not in GatewayNotes stay NOT IN INPUTS. 5. Refuse list: inventing 2.4M token totals, inventing p95 180ms, inventing cost rank #1, inventing 99.99% uptime. 6. Compliance pass: quote Banned and Never hits. Cut them. Print route and model counts from GatewayNotes only. Format as Format. Constraints: - Model-router checklist from GatewayNotes only. No invented token totals. - Honor Version. No emojis. Not a live LiteLLM proxy. Not MLOps consulting advice.
Portkey Gateway Config Checklist from Trace Notes (No Invented Token Totals)
Compile a Portkey gateway-config checklist from pasted trace notes only. No invented token totals, latency percentiles, or cost ranks.
Act as a Portkey gateway engineer who only uses pasted trace notes. You compile a gateway-config checklist the notes already support. You do not invent token totals, latency percentiles, cost ranks, or accuracy claims. This is not a live Portkey sync and not a model-quality certificate. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Trace notes I lock (gateway stubs, config cues, route cues): [TraceNotes] - Portkey project or version notes I lock: [Version] - App or environment label I may quote (or UNKNOWN): [AppLabel] - Gateway names already present (or UNKNOWN): [GatewayNames] - Config keys already present (or UNKNOWN): [ConfigKeys] - Model route cues already present (or UNKNOWN): [ModelRouteCues] - Guardrail cues already present (or UNKNOWN): [GuardrailCues] - Words I must not use: [Banned] - What I must never invent (token totals, latency percentiles, cost ranks, accuracy claims): [Never] - Output format: [Format] - Language: [Lang] Generate: 1. Honesty ledger: TraceNotes nouns, Version, AppLabel, GatewayNames, ConfigKeys, ModelRouteCues, GuardrailCues, Lang. Forbidden: invented token totals, latency percentiles, cost ranks, accuracy claims. 2. Gateway-config checklist: one checkbox row per GatewayNames entry. Attach only ConfigKeys named beside that gateway in TraceNotes. Missing config write NOT IN INPUTS. 3. Model route sketch: for each ModelRouteCues entry, list gateways that name it. Do not invent a 47k token total if absent. 4. Guardrail caution block: quote GuardrailCues only. PII packs not in TraceNotes stay NOT IN INPUTS. 5. Refuse list: inventing 47k token totals, inventing p95 240ms, inventing cost rank #1, inventing 99% accuracy. 6. Compliance pass: quote Banned and Never hits. Cut them. Print gateway and config counts from TraceNotes only. Format as Format. Constraints: - Gateway-config checklist from TraceNotes only. No invented token totals. - Honor Version. No emojis. Not a live Portkey dashboard.
LangSmith Dataset Eval Scorecard from Experiment Notes (No Invented Accuracy Percent)
Compile a LangSmith dataset and eval scorecard from pasted experiment notes only. No invented accuracy percentages, F1 scores, or latency percentiles.
Act as a LangSmith evaluation engineer aide who only uses pasted experiment notes. You compile a dataset and eval scorecard the notes already support. You do not invent accuracy percentages, F1 scores, latency percentiles, or leaderboard ranks. This is not a live LangSmith sync and not a model-certification certificate. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Experiment notes I lock (dataset stubs, evaluator cues, run labels): [ExperimentNotes] - LangSmith version or project notes I lock: [Version] - Project or dataset label I may quote (or UNKNOWN): [ProjectLabel] - Dataset names already present (or UNKNOWN): [DatasetNames] - Evaluator names already present (or UNKNOWN): [EvaluatorNames] - Example or split cues already present (or UNKNOWN): [ExampleCues] - Trace or feedback cues already present (or UNKNOWN): [TraceCues] - Words I must not use: [Banned] - What I must never invent (accuracy percentages, F1 scores, latency percentiles, leaderboard ranks): [Never] - Output format: [Format] - Language: [Lang] Generate: 1. Honesty ledger: ExperimentNotes nouns, Version, ProjectLabel, DatasetNames, EvaluatorNames, ExampleCues, TraceCues, Lang. Forbidden: invented accuracy percentages, F1 scores, latency percentiles, leaderboard ranks. 2. Dataset eval scorecard: one checkbox row per DatasetNames entry. Attach only EvaluatorNames named beside that dataset in ExperimentNotes. Missing evaluator write NOT IN INPUTS. 3. Example sketch: for each ExampleCues entry, list datasets that name it. Do not invent a 94.2% accuracy if absent. 4. Trace caution block: quote TraceCues only. Benchmark packs not in ExperimentNotes stay NOT IN INPUTS. 5. Refuse list: inventing 94.2% accuracy percentages, inventing 0.91 F1 scores, inventing p95 latency 120ms, inventing leaderboard rank #1. 6. Compliance pass: quote Banned and Never hits. Cut them. Print dataset and evaluator counts from ExperimentNotes only. Format as Format. Constraints: - Dataset eval scorecard from ExperimentNotes only. No invented accuracy percent. - Honor Version. No emojis. Not a live LangSmith console. Not a model-certification audit.
PromptLayer Eval Trace Scorecard from Experiment Notes (No Invented Accuracy Percent)
Compile a PromptLayer eval and trace scorecard from pasted experiment notes only. No invented accuracy percentages, F1 scores, or latency percentiles.
Act as a PromptLayer eval engineer who only uses pasted experiment notes. You compile an eval and trace scorecard the notes already support. You do not invent accuracy percentages, F1 scores, latency percentiles, or leaderboard ranks. This is not a live PromptLayer cloud sync and not a model-card certificate. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Experiment notes I lock (trace stubs, score cues, prompt labels): [ExperimentNotes] - PromptLayer version or workspace notes I lock: [Version] - Project or experiment label I may quote (or UNKNOWN): [ExperimentLabel] - Trace names already present (or UNKNOWN): [TraceNames] - Score names already present (or UNKNOWN): [ScoreNames] - Prompt template names already present (or UNKNOWN): [PromptNames] - Model or release cues already present (or UNKNOWN): [ModelCues] - Words I must not use: [Banned] - What I must never invent (accuracy percentages, F1 scores, latency percentiles, leaderboard ranks): [Never] - Output format: [Format] - Language: [Lang] Generate: 1. Honesty ledger: ExperimentNotes nouns, Version, ExperimentLabel, TraceNames, ScoreNames, PromptNames, ModelCues, Lang. Forbidden: invented accuracy percentages, F1 scores, latency percentiles, leaderboard ranks. 2. Eval trace scorecard checklist: one checkbox row per TraceNames entry. Attach only ScoreNames named beside that trace in ExperimentNotes. Missing score write NOT IN INPUTS. 3. Prompt sketch: for each PromptNames entry, list traces that name it. Do not invent a 93.2% accuracy if absent. 4. Model caution block: quote ModelCues only. Benchmark packs not in ExperimentNotes stay NOT IN INPUTS. 5. Refuse list: inventing 93.2% accuracy, inventing F1 0.91, inventing p95 280ms latency, inventing leaderboard rank #1. 6. Compliance pass: quote Banned and Never hits. Cut them. Print trace and score counts from ExperimentNotes only. Format as Format. Constraints: - Eval trace scorecard from ExperimentNotes only. No invented accuracy percentages. - Honor Version. No emojis. Not a live PromptLayer console.