INDEX

Explanations

references to locations, particularly the city of Damascus

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "Damascus"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

owler

-0.91

Trend

-0.81

opsy

-0.78

Scotland

-0.77

ographically

-0.75

ocard

-0.73

inarily

-0.73

INAL

-0.73

rost

-0.73

unal

-0.73

POSITIVE LOGITS

ascus

1.30

 Damascus

1.04

 suburbs

0.86

 rall

0.83

 Assad

0.82

ites

0.81

 suburb

0.77

Assad

0.75

 negoti

0.74

atos

0.74

Activations Density 0.006%

references to locations, particularly the city of Damascus

occurrences of the word "Damascus"

No Comments

No Known Activations