I’m blissful to share that Amazon SageMaker Make clear now helps basis mannequin (FM) analysis (preview). As an information scientist or machine studying (ML) engineer, now you can use SageMaker Make clear to guage, evaluate, and choose FMs in minutes primarily based on metrics similar to accuracy, robustness, creativity, factual data, bias, and toxicity. This new functionality provides to SageMaker Make clear’s current means to detect bias in ML knowledge and fashions and clarify mannequin predictions.

The brand new functionality supplies each computerized and human-in-the-loop evaluations for big language fashions (LLMs) wherever, together with LLMs out there in SageMaker JumpStart, in addition to fashions educated and hosted exterior of AWS. This removes the heavy lifting of discovering the proper mannequin analysis instruments and integrating them into your growth atmosphere. It additionally simplifies the complexity of attempting to undertake educational benchmarks to your generative synthetic intelligence (AI) use case.

Consider FMs with SageMaker Make clear
With SageMaker Make clear, you now have a single place to guage and evaluate any LLM primarily based on predefined standards throughout mannequin choice and all through the mannequin customization workflow. Along with computerized analysis, you can too use the human-in-the-loop capabilities to arrange human evaluations for extra subjective standards, similar to helpfulness, artistic intent, and magnificence, by utilizing your personal workforce or managed workforce from SageMaker Floor Reality.

To get began with mannequin evaluations, you should utilize curated immediate datasets which might be purpose-built for frequent LLM duties, together with open-ended textual content era, textual content summarization, query answering (Q&A), and classification. You may as well lengthen the mannequin analysis with your personal customized immediate datasets and metrics in your particular use case. Human-in-the-loop evaluations can be utilized for any activity and analysis metric. After every analysis job, you obtain an analysis report that summarizes the ends in pure language and consists of visualizations and examples. You possibly can obtain all metrics and stories and in addition combine mannequin evaluations into SageMaker MLOps workflows.

In SageMaker Studio, you will discover Mannequin analysis beneath Jobs within the left menu. You may as well choose Consider immediately from the mannequin particulars web page of any LLM in SageMaker JumpStart.

Choose Consider a mannequin to arrange the analysis job. The UI wizard will information you thru the number of computerized or human analysis, mannequin(s), related duties, metrics, immediate datasets, and overview groups.

As soon as the mannequin analysis job is full, you possibly can view the ends in the analysis report.

Along with the UI, you can too begin with instance Jupyter notebooks that stroll you thru step-by-step directions on the best way to programmatically run mannequin analysis in SageMaker.

Consider fashions wherever with the FMEval open supply library
To run mannequin analysis wherever, together with fashions educated and hosted exterior of AWS, use the FMEval open supply library. The next instance demonstrates the best way to use the library to guage a customized mannequin by extending the ModelRunner class.

For this demo, I select GPT-2 from the Hugging Face mannequin hub and outline a customized HFModelConfig and HuggingFaceCausalLLMModelRunner class that works with causal decoder-only fashions from the Hugging Face mannequin hub similar to GPT-2. The instance can be out there within the FMEval GitHub repo.

!pip set up fmeval

# ModelRunners invoke FMs
from amazon_fmeval.model_runners.model_runner import ModelRunner

# Further imports for customized mannequin
import warnings
from dataclasses import dataclass
from typing import Tuple, Elective
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer

@dataclass
class HFModelConfig:
    model_name: str
    max_new_tokens: int
    normalize_probabilities: bool = False
    seed: int = 0
    remove_prompt_from_generated_text: bool = True

class HuggingFaceCausalLLMModelRunner(ModelRunner):
    def __init__(self, model_config: HFModelConfig):
        self.config = model_config
        self.mannequin = AutoModelForCausalLM.from_pretrained(self.config.model_name)
        self.tokenizer = AutoTokenizer.from_pretrained(self.config.model_name)

    def predict(self, immediate: str) -> Tuple[Optional[str], Elective[float]]:
        input_ids = self.tokenizer(immediate, return_tensors="pt").to(self.mannequin.gadget)
        generations = self.mannequin.generate(
            **input_ids,
            max_new_tokens=self.config.max_new_tokens,
            pad_token_id=self.tokenizer.eos_token_id,
        )
        generation_contains_input = (
            input_ids["input_ids"][0] == generations[0][: input_ids["input_ids"].form[1]]
        ).all()
        if self.config.remove_prompt_from_generated_text and never generation_contains_input:
            warnings.warn(
                "Your mannequin doesn't return the immediate as a part of its generations. "
                "`remove_prompt_from_generated_text` does nothing."
            )
        if self.config.remove_prompt_from_generated_text and generation_contains_input:
            output = self.tokenizer.batch_decode(generations[:, input_ids["input_ids"].form[1] :])[0]
        else:
            output = self.tokenizer.batch_decode(generations, skip_special_tokens=True)[0]

        with torch.inference_mode():
            input_ids = self.tokenizer(self.tokenizer.bos_token + immediate, return_tensors="pt")["input_ids"]
            model_output = self.mannequin(input_ids, labels=input_ids)
            chance = -model_output[0].merchandise()

        return output, chance

Subsequent, create an occasion of HFModelConfig and HuggingFaceCausalLLMModelRunner with the mannequin data.

hf_config = HFModelConfig(model_name="gpt2", max_new_tokens=32)
mannequin = HuggingFaceCausalLLMModelRunner(model_config=hf_config)

Then, choose and configure the analysis algorithm.

# Let's consider the FM for FactualKnowledge
from amazon_fmeval.fmeval import get_eval_algorithm
from amazon_fmeval.eval_algorithms.factual_knowledge import FactualKnowledgeConfig

eval_algorithm_config = FactualKnowledgeConfig("<OR>")
eval_algorithm = get_eval_algorithm("factual_knowledge", eval_algorithm_config)

Let’s first check with one pattern. The analysis rating is the proportion of factually right responses.

model_output = mannequin.predict("London is the capital of")[0]
print(model_output)

eval_algo.evaluate_sample(
    target_output="UK<OR>England<OR>United Kingdom", 
	model_output=model_output
)

the UK, and the UK is the biggest producer of meals on the earth.

The UK is the world's largest producer of meals on the earth.
[EvalScore(name="factual_knowledge", value=1)]

Though it’s not an ideal response, it consists of “UK.”

Subsequent, you possibly can consider the FM utilizing built-in datasets or outline your customized dataset. If you wish to use a customized analysis dataset, create an occasion of DataConfig:

config = DataConfig(
    dataset_name="my_custom_dataset",
    dataset_uri="dataset.jsonl",
    dataset_mime_type=MIME_TYPE_JSONLINES,
    model_input_location="query",
    target_output_location="reply",
)

eval_output = eval_algorithm.consider(
    mannequin=mannequin, 
    dataset_config=config, 
    prompt_template="$function", #$function is changed by the enter worth within the dataset 
    save=True
)

The analysis outcomes will return a mixed analysis rating throughout the dataset and detailed outcomes for every mannequin enter saved in a neighborhood output path.

Be a part of the preview
FM analysis with Amazon SageMaker Make clear is obtainable right now in public preview in AWS Areas US East (Ohio), US East (N. Virginia), US West (Oregon), Asia Pacific (Singapore), Asia Pacific (Tokyo), Europe (Frankfurt), and Europe (Eire). The FMEval open supply library is obtainable on GitHub. To study extra, go to Amazon SageMaker Make clear.

Get began
Log in to the AWS Administration Console and begin evaluating your FMs with SageMaker Make clear right now!

— Antje