INDEX

Explanations

phrases related to a high level of completeness or uniformity

oai_token-act-pair · gpt-3.5-turbo

instances of the word "completely" to emphasize totality or absoluteness

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

maid

-0.86

liest

-0.71

pires

-0.70

rers

-0.69

llor

-0.68

ĺħ

-0.67

hao

-0.66

AMY

-0.66

yip

-0.66

pring

-0.65

POSITIVE LOGITS

BuyableInstoreAndOnline

0.77

 reliant

0.73

 disreg

0.72

 unrelated

0.72

 obliter

0.71

 accomplished

0.69

 depends

0.68

 annihil

0.68

 rewritten

0.67

 overhaul

0.67

Activations Density 0.020%

phrases related to a high level of completeness or uniformity

instances of the word "completely" to emphasize totality or absoluteness

No Comments

No Known Activations