INDEX

Explanations

terms related to questions about future actions or decisions

oai_token-act-pair · gpt-3.5-turbo

references to royalty or high status

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

aniel

-0.78

alos

-0.73

uber

-0.71

byter

-0.69

 Chatt

-0.68

 Celt

-0.67

omo

-0.66

raints

-0.64

isitions

-0.64

ulet

-0.63

POSITIVE LOGITS

PET

0.78

MK

0.68

DP

0.63

pread

0.62

ogical

0.61

 horm

0.61

 Centauri

0.60

NX

0.60

ndra

0.60

IRC

0.60

Activations Density 0.000%

terms related to questions about future actions or decisions

references to royalty or high status

No Comments

No Known Activations