INDEX

Explanations

words related to crime and legal proceedings

oai_token-act-pair · gpt-3.5-turbo

statements about wrongful actions or accusations, particularly in a criminal context

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

Todd

-0.82

 Simmons

-0.77

 Gotham

-0.76

DeL

-0.75

Ohio

-0.73

 Harbor

-0.73

 Garrett

-0.72

 Whitney

-0.72

Chuck

-0.72

 Slate

-0.71

POSITIVE LOGITS

 authorised

0.95

 learnt

0.94

 Malays

0.89

 organis

0.87

 Tamil

0.86

 organise

0.85

 analyse

0.84

jriwal

0.83

 fulfil

0.82

 analys

0.82

Activations Density 3.327%

words related to crime and legal proceedings

statements about wrongful actions or accusations, particularly in a criminal context

No Comments

No Known Activations