INDEX

Explanations

words related to signals or prompts

oai_token-act-pair · gpt-3.5-turbo

references to social cues

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 unsolved

-0.74

ctic

-0.70

olved

-0.68

odder

-0.66

oppable

-0.64

NCT

-0.61

ctica

-0.60

ovation

-0.60

akings

-0.60

aughters

-0.60

POSITIVE LOGITS

cue

1.28

 cues

1.15

llor

0.85

utic

0.83

pill

0.77

vine

0.77

Cue

0.74

ulla

0.72

wcsstore

0.70

xual

0.70

Activations Density 0.008%

words related to signals or prompts

references to social cues

No Comments

No Known Activations