INDEX

Explanations

words related to web links or URLs

oai_token-act-pair · gpt-3.5-turbo

specific substrings or patterns related to the term "iw" in various contexts

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 succeeding

-0.78

âĸ¬

-0.77

inates

-0.65

 fracturing

-0.64

âĸ¬âĸ¬

-0.63

 penetrating

-0.63

Lay

-0.63

 elev

-0.60

 tamp

-0.60

 anat

-0.60

POSITIVE LOGITS

atts

1.13

iw

1.12

ield

1.09

orthy

1.07

ombo

1.06

ugi

1.01

esley

1.01

ielding

0.98

atcher

0.97

esome

0.96

Activations Density 0.006%

words related to web links or URLs

specific substrings or patterns related to the term "iw" in various contexts

No Comments

No Known Activations