INDEX

Explanations

words related to formal statements or discussions

oai_token-act-pair · gpt-3.5-turbo

expressions related to decision-making and opinions

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

 shrouded

-0.67

 inconvenient

-0.65

ulously

-0.64

 swiftly

-0.64

ãĥŁ

-0.63

 stylish

-0.63

");

-0.63

 antic

-0.62

 uncommon

-0.62

 controvers

-0.62

POSITIVE LOGITS

laughs

0.92

plet

0.82

 Dialogue

0.79

 Everybody

0.79

cause

0.78

Yeah

0.75

Laughs

0.74

laughter

0.74

ertodd

0.70

 Alright

0.70

Activations Density 0.893%

words related to formal statements or discussions

expressions related to decision-making and opinions

No Comments

No Known Activations