INDEX

Explanations

words related to specific locations, particularly "Monica" and "Swansea"

oai_token-act-pair · gpt-3.5-turbo

references to specific locations and individual names, particularly "Swansea" and "Monica."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

cles

-0.84

kaya

-0.82

itialized

-0.81

rising

-0.80

WIND

-0.78

rend

-0.76

rant

-0.74

gnu

-0.74

ãĥ´

-0.74

ensibly

-0.74

POSITIVE LOGITS

 Episcopal

0.75

 University

0.67

 Blvd

0.65

mson

0.64

 Pere

0.62

cf

0.62

 Budapest

0.61

 Monstrous

0.61

 Charleston

0.61

 City

0.60

Activations Density 0.033%

words related to specific locations, particularly "Monica" and "Swansea"

references to specific locations and individual names, particularly "Swansea" and "Monica."

No Comments

No Known Activations