INDEX

Explanations

words related to serious negative events or situations

oai_token-act-pair · gpt-3.5-turbo

references to cumulative effects

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

ichick

-0.76

ettel

-0.71

gging

-0.70

rera

-0.66

 Giul

-0.66

 butterflies

-0.66

steamapps

-0.65

ECTION

-0.65

TPPStreamerBot

-0.64

zag

-0.64

POSITIVE LOGITS

ulative

3.00

pg

0.66

dx

0.64

 litter

0.61

volume

0.60

mone

0.59

ularity

0.59

voy

0.58

 Doct

0.58

dor

0.58

Activations Density 0.002%

words related to serious negative events or situations

references to cumulative effects

No Comments

No Known Activations