INDEX

Explanations

terms related to honesty, ethics, and trustworthiness

oai_token-act-pair · gpt-3.5-turbo

discussions surrounding the concept of integrity

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

Adapter

-0.72

joining

-0.70

sg

-0.70

stock

-0.70

Stock

-0.66

agne

-0.65

enegger

-0.64

Pain

-0.64

ergy

-0.63

kson

-0.63

POSITIVE LOGITS

rity

1.19

 integrity

1.07

acies

1.03

 Integrity

0.94

orously

0.79

amental

0.76

ulence

0.74

ately

0.74

 safeguards

0.74

 preservation

0.74

Activations Density 0.028%

terms related to honesty, ethics, and trustworthiness

discussions surrounding the concept of integrity

No Comments

No Known Activations