Upload 3 files

README.md

@@ -1,3 +1,47 @@
 ---
-license: mit
+license: cc0-1.0
+language:
+- en
+task_categories:
+- text-generation
+pretty_name: CompactLM Synthetic Tasks 600
+tags:
+- synthetic
+- instruction-following
+- small-language-models
+- demo
 ---
+
+# CompactLM Synthetic Tasks 600
+
+A **fully synthetic, unofficial demonstration** instruction dataset for experimenting with text-processing pipelines and Hugging Face Datasets. Contains 600 template-based records spanning 20 AI engineering topics and 6 task categories (definition, glossary, summarization, rewriting, extraction, evaluation). There are 540 training records and 60 validation records.
+
+## Important limitations
+
+- All entries were generated using deterministic templates, not sampled from people or real users.
+- Responses are repetitive and have not been independently reviewed. This dataset is **not a benchmark**, a validated training corpus, or evidence of model performance.
+- Train/validation splits can share closely related templates and subjects; **do not treat validation scores as unbiased generalization evidence**.
+- Not affiliated with Hugging Face, NVIDIA, or another organization.
+
+## Format
+
+Each JSONL object contains `id`, `category`, `topic`, `messages` (user and assistant turns), `synthetic`, `source`, and `quality_note`.
+
+## Example
+
+```json
+{"id":"compactlm-0001","category":"definition","topic":"GPU inference","messages":[{"role":"user","content":"Explain GPU inference in one sentence for a beginner."},{"role":"assistant","content":"GPU inference means running a trained model on a graphics processor to produce predictions or outputs."}],"synthetic":true,"source":"template-generated","quality_note":"unreviewed demonstration sample"}
+```
+
+## Loading
+
+```python
+from datasets import load_dataset
+
+dataset = load_dataset("json", data_files={"train": "train.jsonl", "validation": "validation.jsonl"})
+print(dataset["train"][0])
+```
+
+## Intended use
+
+Education, data loading demos, schema testing, and experiments with instruction formatting. Not suitable as a sole training dataset for a production assistant.

validation.jsonl

@@ -0,0 +1,60 @@
+{"id": "compactlm-0436", "category": "rewriting", "topic": "context windows", "messages": [{"role": "user", "content": "Answer concisely: Rewrite for clarity: \"We are doing context windows stuff and checking if it is good, maybe we try bigger later.\""}, {"role": "assistant", "content": "We are evaluating context windows and will consider expanding the experiment after reviewing the results."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0412", "category": "summarization", "topic": "memory usage", "messages": [{"role": "user", "content": "Answer concisely: Summarize this note in a sentence: \"Our team is exploring memory usage. We plan to test it on a small dataset first, measure the result, and document limitations before expanding.\""}, {"role": "assistant", "content": "The team will pilot memory usage, measure outcomes, and document limitations before scaling up."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0470", "category": "evaluation", "topic": "code generation", "messages": [{"role": "user", "content": "Answer concisely: Give two responsible evaluation checks for a demonstration involving code generation."}, {"role": "assistant", "content": "1. Record the test setup and baseline for code generation.\n2. Inspect failures and disclose limitations before drawing conclusions."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0080", "category": "rewriting", "topic": "synthetic datasets", "messages": [{"role": "user", "content": "Rewrite for clarity: \"We are doing synthetic datasets stuff and checking if it is good, maybe we try bigger later.\""}, {"role": "assistant", "content": "We are evaluating synthetic datasets and will consider expanding the experiment after reviewing the results."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0139", "category": "definition", "topic": "classifier calibration", "messages": [{"role": "user", "content": "For a student project: Explain classifier calibration in one sentence for a beginner."}, {"role": "assistant", "content": "Classifier calibration means checking whether predicted probabilities reflect observed outcome frequencies."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0088", "category": "extraction", "topic": "dataset curation", "messages": [{"role": "user", "content": "Extract the focus area and next step as JSON from: \"We are testing dataset curation and will record latency and accuracy.\""}, {"role": "assistant", "content": "{\"focus_area\": \"dataset curation\", \"next_step\": \"record latency and accuracy\"}"}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0551", "category": "rewriting", "topic": "latency optimization", "messages": [{"role": "user", "content": "For technical documentation: Rewrite for clarity: \"We are doing latency optimization stuff and checking if it is good, maybe we try bigger later.\""}, {"role": "assistant", "content": "We are evaluating latency optimization and will consider expanding the experiment after reviewing the results."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0236", "category": "evaluation", "topic": "context windows", "messages": [{"role": "user", "content": "For a student project: Give two responsible evaluation checks for a demonstration involving context windows."}, {"role": "assistant", "content": "1. Record the test setup and baseline for context windows.\n2. Inspect failures and disclose limitations before drawing conclusions."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0405", "category": "summarization", "topic": "quantization", "messages": [{"role": "user", "content": "Answer concisely: Summarize this note in a sentence: \"Our team is exploring quantization. We plan to test it on a small dataset first, measure the result, and document limitations before expanding.\""}, {"role": "assistant", "content": "The team will pilot quantization, measure outcomes, and document limitations before scaling up."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0506", "category": "glossary", "topic": "batching", "messages": [{"role": "user", "content": "For technical documentation: Write a short glossary entry for batching, including one practical consideration."}, {"role": "assistant", "content": "Batching: processing multiple requests together to improve hardware utilization. In practice, evaluate its benefits and trade-offs using measurements from the intended workload."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0023", "category": "glossary", "topic": "tokenization", "messages": [{"role": "user", "content": "Write a short glossary entry for tokenization, including one practical consideration."}, {"role": "assistant", "content": "Tokenization: turning text into smaller units represented by numeric token IDs. In practice, evaluate its benefits and trade-offs using measurements from the intended workload."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0417", "category": "summarization", "topic": "model deployment", "messages": [{"role": "user", "content": "Answer concisely: Summarize this note in a sentence: \"Our team is exploring model deployment. We plan to test it on a small dataset first, measure the result, and document limitations before expanding.\""}, {"role": "assistant", "content": "The team will pilot model deployment, measure outcomes, and document limitations before scaling up."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0096", "category": "extraction", "topic": "context windows", "messages": [{"role": "user", "content": "Extract the focus area and next step as JSON from: \"We are testing context windows and will record latency and accuracy.\""}, {"role": "assistant", "content": "{\"focus_area\": \"context windows\", \"next_step\": \"record latency and accuracy\"}"}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0110", "category": "evaluation", "topic": "code generation", "messages": [{"role": "user", "content": "Give two responsible evaluation checks for a demonstration involving code generation."}, {"role": "assistant", "content": "1. Record the test setup and baseline for code generation.\n2. Inspect failures and disclose limitations before drawing conclusions."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0150", "category": "glossary", "topic": "code generation", "messages": [{"role": "user", "content": "For a student project: Write a short glossary entry for code generation, including one practical consideration."}, {"role": "assistant", "content": "Code generation: producing programming code from natural-language or other input. In practice, evaluate its benefits and trade-offs using measurements from the intended workload."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0410", "category": "summarization", "topic": "code generation", "messages": [{"role": "user", "content": "Answer concisely: Summarize this note in a sentence: \"Our team is exploring code generation. We plan to test it on a small dataset first, measure the result, and document limitations before expanding.\""}, {"role": "assistant", "content": "The team will pilot code generation, measure outcomes, and document limitations before scaling up."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0309", "category": "rewriting", "topic": "prompt evaluation", "messages": [{"role": "user", "content": "For a small research prototype: Rewrite for clarity: \"We are doing prompt evaluation stuff and checking if it is good, maybe we try bigger later.\""}, {"role": "assistant", "content": "We are evaluating prompt evaluation and will consider expanding the experiment after reviewing the results."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0059", "category": "summarization", "topic": "classifier calibration", "messages": [{"role": "user", "content": "Summarize this note in a sentence: \"Our team is exploring classifier calibration. We plan to test it on a small dataset first, measure the result, and document limitations before expanding.\""}, {"role": "assistant", "content": "The team will pilot classifier calibration, measure outcomes, and document limitations before scaling up."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0446", "category": "extraction", "topic": "batching", "messages": [{"role": "user", "content": "Answer concisely: Extract the focus area and next step as JSON from: \"We are testing batching and will record latency and accuracy.\""}, {"role": "assistant", "content": "{\"focus_area\": \"batching\", \"next_step\": \"record latency and accuracy\"}"}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0187", "category": "rewriting", "topic": "model distillation", "messages": [{"role": "user", "content": "For a student project: Rewrite for clarity: \"We are doing model distillation stuff and checking if it is good, maybe we try bigger later.\""}, {"role": "assistant", "content": "We are evaluating model distillation and will consider expanding the experiment after reviewing the results."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0027", "category": "glossary", "topic": "model distillation", "messages": [{"role": "user", "content": "Write a short glossary entry for model distillation, including one practical consideration."}, {"role": "assistant", "content": "Model distillation: training a smaller student model using guidance from a stronger teacher. In practice, evaluate its benefits and trade-offs using measurements from the intended workload."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0231", "category": "evaluation", "topic": "latency optimization", "messages": [{"role": "user", "content": "For a student project: Give two responsible evaluation checks for a demonstration involving latency optimization."}, {"role": "assistant", "content": "1. Record the test setup and baseline for latency optimization.\n2. Inspect failures and disclose limitations before drawing conclusions."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0326", "category": "extraction", "topic": "batching", "messages": [{"role": "user", "content": "For a small research prototype: Extract the focus area and next step as JSON from: \"We are testing batching and will record latency and accuracy.\""}, {"role": "assistant", "content": "{\"focus_area\": \"batching\", \"next_step\": \"record latency and accuracy\"}"}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0273", "category": "glossary", "topic": "retrieval augmentation", "messages": [{"role": "user", "content": "For a small research prototype: Write a short glossary entry for retrieval augmentation, including one practical consideration."}, {"role": "assistant", "content": "Retrieval augmentation: supplementing generation with information retrieved from an external collection. In practice, evaluate its benefits and trade-offs using measurements from the intended workload."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0055", "category": "summarization", "topic": "training validation", "messages": [{"role": "user", "content": "Summarize this note in a sentence: \"Our team is exploring training validation. We plan to test it on a small dataset first, measure the result, and document limitations before expanding.\""}, {"role": "assistant", "content": "The team will pilot training validation, measure outcomes, and document limitations before scaling up."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0183", "category": "rewriting", "topic": "tokenization", "messages": [{"role": "user", "content": "For a student project: Rewrite for clarity: \"We are doing tokenization stuff and checking if it is good, maybe we try bigger later.\""}, {"role": "assistant", "content": "We are evaluating tokenization and will consider expanding the experiment after reviewing the results."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0148", "category": "glossary", "topic": "dataset curation", "messages": [{"role": "user", "content": "For a student project: Write a short glossary entry for dataset curation, including one practical consideration."}, {"role": "assistant", "content": "Dataset curation: selecting, cleaning, and documenting examples used for training or evaluation. In practice, evaluate its benefits and trade-offs using measurements from the intended workload."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0102", "category": "evaluation", "topic": "on-device AI", "messages": [{"role": "user", "content": "Give two responsible evaluation checks for a demonstration involving on-device AI."}, {"role": "assistant", "content": "1. Record the test setup and baseline for on-device AI.\n2. Inspect failures and disclose limitations before drawing conclusions."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0127", "category": "definition", "topic": "model distillation", "messages": [{"role": "user", "content": "For a student project: Explain model distillation in one sentence for a beginner."}, {"role": "assistant", "content": "Model distillation means training a smaller student model using guidance from a stronger teacher."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0588", "category": "evaluation", "topic": "dataset curation", "messages": [{"role": "user", "content": "For technical documentation: Give two responsible evaluation checks for a demonstration involving dataset curation."}, {"role": "assistant", "content": "1. Record the test setup and baseline for dataset curation.\n2. Inspect failures and disclose limitations before drawing conclusions."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0569", "category": "extraction", "topic": "prompt evaluation", "messages": [{"role": "user", "content": "For technical documentation: Extract the focus area and next step as JSON from: \"We are testing prompt evaluation and will record latency and accuracy.\""}, {"role": "assistant", "content": "{\"focus_area\": \"prompt evaluation\", \"next_step\": \"record latency and accuracy\"}"}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0129", "category": "definition", "topic": "prompt evaluation", "messages": [{"role": "user", "content": "For a student project: Explain prompt evaluation in one sentence for a beginner."}, {"role": "assistant", "content": "Prompt evaluation means testing how model outputs change in response to prompts."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0487", "category": "definition", "topic": "model distillation", "messages": [{"role": "user", "content": "For technical documentation: Explain model distillation in one sentence for a beginner."}, {"role": "assistant", "content": "Model distillation means training a smaller student model using guidance from a stronger teacher."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0369", "category": "definition", "topic": "prompt evaluation", "messages": [{"role": "user", "content": "Answer concisely: Explain prompt evaluation in one sentence for a beginner."}, {"role": "assistant", "content": "Prompt evaluation means testing how model outputs change in response to prompts."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0075", "category": "rewriting", "topic": "training validation", "messages": [{"role": "user", "content": "Rewrite for clarity: \"We are doing training validation stuff and checking if it is good, maybe we try bigger later.\""}, {"role": "assistant", "content": "We are evaluating training validation and will consider expanding the experiment after reviewing the results."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0222", "category": "evaluation", "topic": "on-device AI", "messages": [{"role": "user", "content": "For a student project: Give two responsible evaluation checks for a demonstration involving on-device AI."}, {"role": "assistant", "content": "1. Record the test setup and baseline for on-device AI.\n2. Inspect failures and disclose limitations before drawing conclusions."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0289", "category": "summarization", "topic": "prompt evaluation", "messages": [{"role": "user", "content": "For a small research prototype: Summarize this note in a sentence: \"Our team is exploring prompt evaluation. We plan to test it on a small dataset first, measure the result, and document limitations before expanding.\""}, {"role": "assistant", "content": "The team will pilot prompt evaluation, measure outcomes, and document limitations before scaling up."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0443", "category": "extraction", "topic": "tokenization", "messages": [{"role": "user", "content": "Answer concisely: Extract the focus area and next step as JSON from: \"We are testing tokenization and will record latency and accuracy.\""}, {"role": "assistant", "content": "{\"focus_area\": \"tokenization\", \"next_step\": \"record latency and accuracy\"}"}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0034", "category": "glossary", "topic": "fine-tuning", "messages": [{"role": "user", "content": "Write a short glossary entry for fine-tuning, including one practical consideration."}, {"role": "assistant", "content": "Fine-tuning: continuing model training on a smaller task- or domain-specific dataset. In practice, evaluate its benefits and trade-offs using measurements from the intended workload."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0123", "category": "definition", "topic": "tokenization", "messages": [{"role": "user", "content": "For a student project: Explain tokenization in one sentence for a beginner."}, {"role": "assistant", "content": "Tokenization means turning text into smaller units represented by numeric token IDs."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0061", "category": "rewriting", "topic": "GPU inference", "messages": [{"role": "user", "content": "Rewrite for clarity: \"We are doing GPU inference stuff and checking if it is good, maybe we try bigger later.\""}, {"role": "assistant", "content": "We are evaluating GPU inference and will consider expanding the experiment after reviewing the results."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0264", "category": "glossary", "topic": "embedding search", "messages": [{"role": "user", "content": "For a small research prototype: Write a short glossary entry for embedding search, including one practical consideration."}, {"role": "assistant", "content": "Embedding search: retrieving items using similarity between numerical vector representations. In practice, evaluate its benefits and trade-offs using measurements from the intended workload."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0172", "category": "summarization", "topic": "memory usage", "messages": [{"role": "user", "content": "For a student project: Summarize this note in a sentence: \"Our team is exploring memory usage. We plan to test it on a small dataset first, measure the result, and document limitations before expanding.\""}, {"role": "assistant", "content": "The team will pilot memory usage, measure outcomes, and document limitations before scaling up."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0327", "category": "extraction", "topic": "model distillation", "messages": [{"role": "user", "content": "For a small research prototype: Extract the focus area and next step as JSON from: \"We are testing model distillation and will record latency and accuracy.\""}, {"role": "assistant", "content": "{\"focus_area\": \"model distillation\", \"next_step\": \"record latency and accuracy\"}"}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0154", "category": "glossary", "topic": "fine-tuning", "messages": [{"role": "user", "content": "For a student project: Write a short glossary entry for fine-tuning, including one practical consideration."}, {"role": "assistant", "content": "Fine-tuning: continuing model training on a smaller task- or domain-specific dataset. In practice, evaluate its benefits and trade-offs using measurements from the intended workload."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0159", "category": "glossary", "topic": "classifier calibration", "messages": [{"role": "user", "content": "For a student project: Write a short glossary entry for classifier calibration, including one practical consideration."}, {"role": "assistant", "content": "Classifier calibration: checking whether predicted probabilities reflect observed outcome frequencies. In practice, evaluate its benefits and trade-offs using measurements from the intended workload."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0450", "category": "extraction", "topic": "code generation", "messages": [{"role": "user", "content": "Answer concisely: Extract the focus area and next step as JSON from: \"We are testing code generation and will record latency and accuracy.\""}, {"role": "assistant", "content": "{\"focus_area\": \"code generation\", \"next_step\": \"record latency and accuracy\"}"}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0019", "category": "definition", "topic": "classifier calibration", "messages": [{"role": "user", "content": "Explain classifier calibration in one sentence for a beginner."}, {"role": "assistant", "content": "Classifier calibration means checking whether predicted probabilities reflect observed outcome frequencies."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0256", "category": "definition", "topic": "context windows", "messages": [{"role": "user", "content": "For a small research prototype: Explain context windows in one sentence for a beginner."}, {"role": "assistant", "content": "Context windows means the limit on how many input and generated tokens a model can consider in one request."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0010", "category": "definition", "topic": "code generation", "messages": [{"role": "user", "content": "Explain code generation in one sentence for a beginner."}, {"role": "assistant", "content": "Code generation means producing programming code from natural-language or other input."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0393", "category": "glossary", "topic": "retrieval augmentation", "messages": [{"role": "user", "content": "Answer concisely: Write a short glossary entry for retrieval augmentation, including one practical consideration."}, {"role": "assistant", "content": "Retrieval augmentation: supplementing generation with information retrieved from an external collection. In practice, evaluate its benefits and trade-offs using measurements from the intended workload."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0284", "category": "summarization", "topic": "embedding search", "messages": [{"role": "user", "content": "For a small research prototype: Summarize this note in a sentence: \"Our team is exploring embedding search. We plan to test it on a small dataset first, measure the result, and document limitations before expanding.\""}, {"role": "assistant", "content": "The team will pilot embedding search, measure outcomes, and document limitations before scaling up."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0567", "category": "extraction", "topic": "model distillation", "messages": [{"role": "user", "content": "For technical documentation: Extract the focus area and next step as JSON from: \"We are testing model distillation and will record latency and accuracy.\""}, {"role": "assistant", "content": "{\"focus_area\": \"model distillation\", \"next_step\": \"record latency and accuracy\"}"}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0291", "category": "summarization", "topic": "latency optimization", "messages": [{"role": "user", "content": "For a small research prototype: Summarize this note in a sentence: \"Our team is exploring latency optimization. We plan to test it on a small dataset first, measure the result, and document limitations before expanding.\""}, {"role": "assistant", "content": "The team will pilot latency optimization, measure outcomes, and document limitations before scaling up."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0592", "category": "evaluation", "topic": "memory usage", "messages": [{"role": "user", "content": "For technical documentation: Give two responsible evaluation checks for a demonstration involving memory usage."}, {"role": "assistant", "content": "1. Record the test setup and baseline for memory usage.\n2. Inspect failures and disclose limitations before drawing conclusions."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0395", "category": "glossary", "topic": "training validation", "messages": [{"role": "user", "content": "Answer concisely: Write a short glossary entry for training validation, including one practical consideration."}, {"role": "assistant", "content": "Training validation: using held-out examples to assess model behavior during development. In practice, evaluate its benefits and trade-offs using measurements from the intended workload."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0171", "category": "summarization", "topic": "latency optimization", "messages": [{"role": "user", "content": "For a student project: Summarize this note in a sentence: \"Our team is exploring latency optimization. We plan to test it on a small dataset first, measure the result, and document limitations before expanding.\""}, {"role": "assistant", "content": "The team will pilot latency optimization, measure outcomes, and document limitations before scaling up."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0237", "category": "evaluation", "topic": "model deployment", "messages": [{"role": "user", "content": "For a student project: Give two responsible evaluation checks for a demonstration involving model deployment."}, {"role": "assistant", "content": "1. Record the test setup and baseline for model deployment.\n2. Inspect failures and disclose limitations before drawing conclusions."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0341", "category": "evaluation", "topic": "GPU inference", "messages": [{"role": "user", "content": "For a small research prototype: Give two responsible evaluation checks for a demonstration involving GPU inference."}, {"role": "assistant", "content": "1. Record the test setup and baseline for GPU inference.\n2. Inspect failures and disclose limitations before drawing conclusions."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}
+{"id": "compactlm-0391", "category": "glossary", "topic": "latency optimization", "messages": [{"role": "user", "content": "Answer concisely: Write a short glossary entry for latency optimization, including one practical consideration."}, {"role": "assistant", "content": "Latency optimization: reducing the delay between an input request and a generated response. In practice, evaluate its benefits and trade-offs using measurements from the intended workload."}], "synthetic": true, "source": "template-generated", "quality_note": "unreviewed demonstration sample"}