INDEX

Explanations

the word "escape"

oai_token-act-pair · gpt-3.5-turbo

mentions of events or situations that could happen

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

redits

-0.85

amins

-0.79

âĶĢâĶĢâĶĢâĶĢ

-0.74

ependent

-0.73

rift

-0.72

imester

-0.71

itamin

-0.71

ompl

-0.70

iability

-0.68

orb

-0.67

POSITIVE LOGITS

llah

0.79

 Canad

0.72

 Lancaster

0.66

fred

0.66

 Johnston

0.65

 Amos

0.65

 Bing

0.65

 Romeo

0.65

Iro

0.64

feel

0.64

Activations Density 0.000%

the word "escape"

mentions of events or situations that could happen

No Comments

No Known Activations