INDEX

Explanations

adjectives or nouns indicating importance, supremacy, or priority

oai_token-act-pair · gpt-3.5-turbo

terms that indicate primary or dominant roles and influences

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

oops

-0.85

erved

-0.81

zanne

-0.79

earances

-0.76

ancies

-0.73

ravings

-0.72

undreds

-0.72

atri

-0.71

ires

-0.69

IRE

-0.68

POSITIVE LOGITS

 beneficiary

1.19

 conduit

1.11

 culprit

1.06

 source

1.03

 obstacle

0.99

 indicator

0.98

 casualty

0.97

 catalyst

0.97

 contender

0.94

 motiv

0.93

Activations Density 0.171%

adjectives or nouns indicating importance, supremacy, or priority

terms that indicate primary or dominant roles and influences

No Comments

No Known Activations