Samples
5
Total tokens
-
not reported by provider
Latency (avg)
0.00 ms
median 0.00 ms max 0.00 ms
Summary
| Metric | Score | Distribution | Threshold | Status |
|---|---|---|---|---|
| exact_match | 0.400 | >= 0.500 | FAIL | |
| hallucination_flag | 0.900 | >= 0.850 | PASS |
Samples (5)
#0What does LLM stand for?1.001.00
Input
What does LLM stand for?
Expected
Large Language Model
Context (1)
[0]A Large Language Model (LLM) is a type of neural network trained on vast amounts of text data.
Response
Large Language Model
Metrics
| Metric | Score | Reason |
|---|---|---|
| exact_match | 1.000 | match |
| hallucination_flag | 1.000 | All extracted tokens found in context. |
#1Who introduced the Transformer architecture?0.000.50
Input
Who introduced the Transformer architecture?
Expected
Context (1)
[0]The Transformer architecture was introduced in the paper 'Attention is All You Need' by researchers at Google in 2017.
Response
OpenAI in 2017
Metrics
| Metric | Score | Reason |
|---|---|---|
| exact_match | 0.000 | mismatch |
| hallucination_flag | 0.500 | 1/2 tokens not found in context. |
#2When was BERT first released?0.001.00
Input
When was BERT first released?
Expected
2018
Context (1)
[0]BERT was released by Google in October 2018, becoming a foundational model for downstream NLP tasks.
Response
October 2018
Metrics
| Metric | Score | Reason |
|---|---|---|
| exact_match | 0.000 | mismatch |
| hallucination_flag | 1.000 | All extracted tokens found in context. |
#3What does the temperature parameter control in an LLM?0.001.00
Input
What does the temperature parameter control in an LLM?
Expected
Controls randomness
Context (1)
[0]Temperature is a hyperparameter that controls the randomness of LLM outputs. Lower values yield more deterministic responses.
Response
Controls randomness in output
Metrics
| Metric | Score | Reason |
|---|---|---|
| exact_match | 0.000 | mismatch |
| hallucination_flag | 1.000 | All extracted tokens found in context. |
#4What does RAG stand for?1.001.00
Input
What does RAG stand for?
Expected
Retrieval-Augmented Generation
Context (1)
[0]RAG (Retrieval-Augmented Generation) combines retrieval of relevant documents with text generation to ground LLM responses in external knowledge.
Response
Retrieval-Augmented Generation
Metrics
| Metric | Score | Reason |
|---|---|---|
| exact_match | 1.000 | match |
| hallucination_flag | 1.000 | All extracted tokens found in context. |
No samples match the current filter.