INDEX

Explanations

words related to legal terms, specifically the concept of guilt

oai_token-act-pair · gpt-3.5-turbo

instances of the word "guilty" in the context of legal judgments or accusations

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

andel

-0.79

ILA

-0.75

pid

-0.73

flies

-0.73

edia

-0.72

yip

-0.72

abwe

-0.69

 Gork

-0.68

pora

-0.67

lav

-0.67

POSITIVE LOGITS

 plea

0.86

 verdict

0.85

 Guilty

0.84

 pleas

0.83

 guilty

0.80

isance

0.78

 unfocusedRange

0.76

alty

0.76

thouse

0.74

alties

0.73

Activations Density 0.015%

words related to legal terms, specifically the concept of guilt

instances of the word "guilty" in the context of legal judgments or accusations

No Comments

No Known Activations