INDEX

Explanations

superlatives or extremes, such as "ultimate", "new", or "perfect"

oai_token-act-pair · gpt-3.5-turbo

terms that denote superiority or excellence

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

hire

-0.85

rehend

-0.80

imester

-0.80

related

-0.78

attr

-0.76

ription

-0.76

laden

-0.76

VIEW

-0.75

cies

-0.74

onis

-0.74

POSITIVE LOGITS

 antidote

0.93

 arbit

0.87

 embodiment

0.87

 underdog

0.85

 culprit

0.85

 conduit

0.83

 catalyst

0.83

 beneficiary

0.83

 gateway

0.82

 Trojan

0.81

Activations Density 0.252%

superlatives or extremes, such as "ultimate", "new", or "perfect"

terms that denote superiority or excellence

No Comments

No Known Activations