Evaluating LLMs: beyond vibes
Building eval suites that matter. This post digs into the topic, drawing on recent discussion from Hacker News, Reddit, and the wider community.
The conversation that prompted this post is linked in the sources. I have tried to capture the substance of the discussion rather than just summarize it, because the comments are usually where the real insight lives.
What happened
Fine-tuning is overused. Most teams should exhaust prompting and RAG before considering a fine-tune, because fine-tuning is expensive to evaluate and hard to iterate on.
Observability for LLM apps means logging prompts, completions, tool calls, and costs. Without it you are flying blind.
Latency is a feature. Users abandon slow AI features faster than slow traditional ones because the expectation of instant response is higher.
Context windows keep growing, but attention is not free. Stuffing more tokens into the prompt dilutes the signal and increases cost and latency. Retrieval and summarization usually beat brute force.
Multi-agent systems are promising but add coordination complexity. Start with a single agent and a good tool set before splitting into specialists.
Latency is a feature. Users abandon slow AI features faster than slow traditional ones because the expectation of instant response is higher.
Context windows keep growing, but attention is not free. Stuffing more tokens into the prompt dilutes the signal and increases cost and latency. Retrieval and summarization usually beat brute force.
Guardrails are not optional in production. Models will eventually produce something you did not want, and the question is whether your system catches it before the user does.
Model routing sends simple queries to cheap models and hard ones to expensive models. It is the easiest cost optimization most teams have not done.
Fine-tuning is overused. Most teams should exhaust prompting and RAG before considering a fine-tune, because fine-tuning is expensive to evaluate and hard to iterate on.
Fine-tuning is overused. Most teams should exhaust prompting and RAG before considering a fine-tune, because fine-tuning is expensive to evaluate and hard to iterate on.
Tool use turns a chat model into an agent. The model decides which tool to call, with what arguments, and interprets the result. Getting this loop right is where most of the engineering effort goes.
Fine-tuning is overused. Most teams should exhaust prompting and RAG before considering a fine-tune, because fine-tuning is expensive to evaluate and hard to iterate on.
Multi-agent systems are promising but add coordination complexity. Start with a single agent and a good tool set before splitting into specialists.
Prompt engineering is real engineering. The difference between a mediocre and an excellent prompt is often the difference between a demo and a product, and it deserves the same rigor as code.
Why it matters
Guardrails are not optional in production. Models will eventually produce something you did not want, and the question is whether your system catches it before the user does.
Tool use turns a chat model into an agent. The model decides which tool to call, with what arguments, and interprets the result. Getting this loop right is where most of the engineering effort goes.
Observability for LLM apps means logging prompts, completions, tool calls, and costs. Without it you are flying blind.
Evaluation is a product, not a test. A good eval suite encodes what your users actually care about and becomes the spec your model is measured against.
Multi-agent systems are promising but add coordination complexity. Start with a single agent and a good tool set before splitting into specialists.
Latency is a feature. Users abandon slow AI features faster than slow traditional ones because the expectation of instant response is higher.
Observability for LLM apps means logging prompts, completions, tool calls, and costs. Without it you are flying blind.
Prompt engineering is real engineering. The difference between a mediocre and an excellent prompt is often the difference between a demo and a product, and it deserves the same rigor as code.
Model routing sends simple queries to cheap models and hard ones to expensive models. It is the easiest cost optimization most teams have not done.
The hardest part of building with LLMs is not the model call. It is the evaluation. Without a solid eval suite, you cannot tell whether a change helped or hurt, and you end up shipping on vibes.
Context windows keep growing, but attention is not free. Stuffing more tokens into the prompt dilutes the signal and increases cost and latency. Retrieval and summarization usually beat brute force.
Structured output via JSON schema constraints makes LLMs usable as components in larger systems. Unstructured text is a liability in a pipeline.
Structured output via JSON schema constraints makes LLMs usable as components in larger systems. Unstructured text is a liability in a pipeline.
Model routing sends simple queries to cheap models and hard ones to expensive models. It is the easiest cost optimization most teams have not done.
Guardrails are not optional in production. Models will eventually produce something you did not want, and the question is whether your system catches it before the user does.
Practical takeaways
- Caching is the single biggest cost lever for LLM applications.
- Observability for LLM apps means logging prompts, completions, tool calls, and costs.
- Structured output via JSON schema constraints makes LLMs usable as components in larger systems.
- Multi-agent systems are promising but add coordination complexity.
- Context windows keep growing, but attention is not free.
- Model routing sends simple queries to cheap models and hard ones to expensive models.
- Large language models are best understood as very good pattern matchers with a large but finite context.
- Context windows keep growing, but attention is not free.
- Latency is a feature.
- Context windows keep growing, but attention is not free.
Model routing sends simple queries to cheap models and hard ones to expensive models. It is the easiest cost optimization most teams have not done.
Evaluation is a product, not a test. A good eval suite encodes what your users actually care about and becomes the spec your model is measured against.
Evaluation is a product, not a test. A good eval suite encodes what your users actually care about and becomes the spec your model is measured against.
Retrieval quality dominates RAG quality. A perfect model with bad retrieval gives bad answers, and a mediocre model with great retrieval gives acceptable ones.
Fine-tuning is overused. Most teams should exhaust prompting and RAG before considering a fine-tune, because fine-tuning is expensive to evaluate and hard to iterate on.
Model routing sends simple queries to cheap models and hard ones to expensive models. It is the easiest cost optimization most teams have not done.
Prompt engineering is real engineering. The difference between a mediocre and an excellent prompt is often the difference between a demo and a product, and it deserves the same rigor as code.
Fine-tuning is overused. Most teams should exhaust prompting and RAG before considering a fine-tune, because fine-tuning is expensive to evaluate and hard to iterate on.
Multi-agent systems are promising but add coordination complexity. Start with a single agent and a good tool set before splitting into specialists.
Fine-tuning is overused. Most teams should exhaust prompting and RAG before considering a fine-tune, because fine-tuning is expensive to evaluate and hard to iterate on.
Caching is the single biggest cost lever for LLM applications. Prompt caching and response caching can cut bills by an order of magnitude for workloads with repeated prefixes.
Caching is the single biggest cost lever for LLM applications. Prompt caching and response caching can cut bills by an order of magnitude for workloads with repeated prefixes.
Multi-agent systems are promising but add coordination complexity. Start with a single agent and a good tool set before splitting into specialists.
Structured output via JSON schema constraints makes LLMs usable as components in larger systems. Unstructured text is a liability in a pipeline.
Multi-agent systems are promising but add coordination complexity. Start with a single agent and a good tool set before splitting into specialists.
Guardrails are not optional in production. Models will eventually produce something you did not want, and the question is whether your system catches it before the user does.
Fine-tuning is overused. Most teams should exhaust prompting and RAG before considering a fine-tune, because fine-tuning is expensive to evaluate and hard to iterate on.
Observability for LLM apps means logging prompts, completions, tool calls, and costs. Without it you are flying blind.
Caching is the single biggest cost lever for LLM applications. Prompt caching and response caching can cut bills by an order of magnitude for workloads with repeated prefixes.
Model routing sends simple queries to cheap models and hard ones to expensive models. It is the easiest cost optimization most teams have not done.
Sources
This post is informed by discussion on hamel.dev and related threads on Hacker News and Reddit. Primary source: https://hamel.dev/blog/posts/evals/.
Observability for LLM apps means logging prompts, completions, tool calls, and costs. Without it you are flying blind.
Evaluation is a product, not a test. A good eval suite encodes what your users actually care about and becomes the spec your model is measured against.
Context windows keep growing, but attention is not free. Stuffing more tokens into the prompt dilutes the signal and increases cost and latency. Retrieval and summarization usually beat brute force.
Structured output via JSON schema constraints makes LLMs usable as components in larger systems. Unstructured text is a liability in a pipeline.
Large language models are best understood as very good pattern matchers with a large but finite context. They are not databases, not search engines, and not oracles, and treating them as any of those leads to predictable failures.
Multi-agent systems are promising but add coordination complexity. Start with a single agent and a good tool set before splitting into specialists.
Fine-tuning is overused. Most teams should exhaust prompting and RAG before considering a fine-tune, because fine-tuning is expensive to evaluate and hard to iterate on.
Context windows keep growing, but attention is not free. Stuffing more tokens into the prompt dilutes the signal and increases cost and latency. Retrieval and summarization usually beat brute force.
Large language models are best understood as very good pattern matchers with a large but finite context. They are not databases, not search engines, and not oracles, and treating them as any of those leads to predictable failures.
Model routing sends simple queries to cheap models and hard ones to expensive models. It is the easiest cost optimization most teams have not done.
Caching is the single biggest cost lever for LLM applications. Prompt caching and response caching can cut bills by an order of magnitude for workloads with repeated prefixes.
Large language models are best understood as very good pattern matchers with a large but finite context. They are not databases, not search engines, and not oracles, and treating them as any of those leads to predictable failures.
Structured output via JSON schema constraints makes LLMs usable as components in larger systems. Unstructured text is a liability in a pipeline.
Observability for LLM apps means logging prompts, completions, tool calls, and costs. Without it you are flying blind.
Guardrails are not optional in production. Models will eventually produce something you did not want, and the question is whether your system catches it before the user does.
Structured output via JSON schema constraints makes LLMs usable as components in larger systems. Unstructured text is a liability in a pipeline.
Prompt engineering is real engineering. The difference between a mediocre and an excellent prompt is often the difference between a demo and a product, and it deserves the same rigor as code.
Evaluation is a product, not a test. A good eval suite encodes what your users actually care about and becomes the spec your model is measured against.
Multi-agent systems are promising but add coordination complexity. Start with a single agent and a good tool set before splitting into specialists.
Context windows keep growing, but attention is not free. Stuffing more tokens into the prompt dilutes the signal and increases cost and latency. Retrieval and summarization usually beat brute force.
Multi-agent systems are promising but add coordination complexity. Start with a single agent and a good tool set before splitting into specialists.
The hardest part of building with LLMs is not the model call. It is the evaluation. Without a solid eval suite, you cannot tell whether a change helped or hurt, and you end up shipping on vibes.
Prompt engineering is real engineering. The difference between a mediocre and an excellent prompt is often the difference between a demo and a product, and it deserves the same rigor as code.
Guardrails are not optional in production. Models will eventually produce something you did not want, and the question is whether your system catches it before the user does.
Structured output via JSON schema constraints makes LLMs usable as components in larger systems. Unstructured text is a liability in a pipeline.
Multi-agent systems are promising but add coordination complexity. Start with a single agent and a good tool set before splitting into specialists.
Model routing sends simple queries to cheap models and hard ones to expensive models. It is the easiest cost optimization most teams have not done.
Evaluation is a product, not a test. A good eval suite encodes what your users actually care about and becomes the spec your model is measured against.
Evaluation is a product, not a test. A good eval suite encodes what your users actually care about and becomes the spec your model is measured against.
Prompt engineering is real engineering. The difference between a mediocre and an excellent prompt is often the difference between a demo and a product, and it deserves the same rigor as code.
Multi-agent systems are promising but add coordination complexity. Start with a single agent and a good tool set before splitting into specialists.
Caching is the single biggest cost lever for LLM applications. Prompt caching and response caching can cut bills by an order of magnitude for workloads with repeated prefixes.
Tool use turns a chat model into an agent. The model decides which tool to call, with what arguments, and interprets the result. Getting this loop right is where most of the engineering effort goes.
Prompt engineering is real engineering. The difference between a mediocre and an excellent prompt is often the difference between a demo and a product, and it deserves the same rigor as code.
Model routing sends simple queries to cheap models and hard ones to expensive models. It is the easiest cost optimization most teams have not done.
Large language models are best understood as very good pattern matchers with a large but finite context. They are not databases, not search engines, and not oracles, and treating them as any of those leads to predictable failures.
Structured output via JSON schema constraints makes LLMs usable as components in larger systems. Unstructured text is a liability in a pipeline.
Multi-agent systems are promising but add coordination complexity. Start with a single agent and a good tool set before splitting into specialists.
The hardest part of building with LLMs is not the model call. It is the evaluation. Without a solid eval suite, you cannot tell whether a change helped or hurt, and you end up shipping on vibes.
Fine-tuning is overused. Most teams should exhaust prompting and RAG before considering a fine-tune, because fine-tuning is expensive to evaluate and hard to iterate on.
The hardest part of building with LLMs is not the model call. It is the evaluation. Without a solid eval suite, you cannot tell whether a change helped or hurt, and you end up shipping on vibes.
Evaluation is a product, not a test. A good eval suite encodes what your users actually care about and becomes the spec your model is measured against.
Observability for LLM apps means logging prompts, completions, tool calls, and costs. Without it you are flying blind.
Guardrails are not optional in production. Models will eventually produce something you did not want, and the question is whether your system catches it before the user does.
Context windows keep growing, but attention is not free. Stuffing more tokens into the prompt dilutes the signal and increases cost and latency. Retrieval and summarization usually beat brute force.
Retrieval quality dominates RAG quality. A perfect model with bad retrieval gives bad answers, and a mediocre model with great retrieval gives acceptable ones.
Multi-agent systems are promising but add coordination complexity. Start with a single agent and a good tool set before splitting into specialists.
Observability for LLM apps means logging prompts, completions, tool calls, and costs. Without it you are flying blind.
Guardrails are not optional in production. Models will eventually produce something you did not want, and the question is whether your system catches it before the user does.
Structured output via JSON schema constraints makes LLMs usable as components in larger systems. Unstructured text is a liability in a pipeline.
Evaluation is a product, not a test. A good eval suite encodes what your users actually care about and becomes the spec your model is measured against.
Observability for LLM apps means logging prompts, completions, tool calls, and costs. Without it you are flying blind.
Multi-agent systems are promising but add coordination complexity. Start with a single agent and a good tool set before splitting into specialists.
Latency is a feature. Users abandon slow AI features faster than slow traditional ones because the expectation of instant response is higher.
Fine-tuning is overused. Most teams should exhaust prompting and RAG before considering a fine-tune, because fine-tuning is expensive to evaluate and hard to iterate on.
Large language models are best understood as very good pattern matchers with a large but finite context. They are not databases, not search engines, and not oracles, and treating them as any of those leads to predictable failures.
Evaluation is a product, not a test. A good eval suite encodes what your users actually care about and becomes the spec your model is measured against.
The hardest part of building with LLMs is not the model call. It is the evaluation. Without a solid eval suite, you cannot tell whether a change helped or hurt, and you end up shipping on vibes.
Fine-tuning is overused. Most teams should exhaust prompting and RAG before considering a fine-tune, because fine-tuning is expensive to evaluate and hard to iterate on.
Evaluation is a product, not a test. A good eval suite encodes what your users actually care about and becomes the spec your model is measured against.