INDEX

Explanations

names of people or places

oai_token-act-pair · gpt-3.5-turbo

references to specific names or entities

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

ober

-0.70

ively

-0.68

lich

-0.67

okin

-0.67

 wrench

-0.66

ICES

-0.65

enegger

-0.64

ivity

-0.63

OPT

-0.62

ysis

-0.62

POSITIVE LOGITS

zza

0.87

geon

0.76

 Cheong

0.75

nesday

0.74

peria

0.73

ãĥ¼ãĤ¯

0.71

hiro

0.71

qua

0.70

æ©

0.70

venture

0.70

Activations Density 0.050%

names of people or places

references to specific names or entities

No Comments

No Known Activations