INDEX

Explanations

dates, names, and locations typically seen in news articles

oai_token-act-pair · gpt-3.5-turbo

geographical locations and proper nouns

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

 Reincarn

-0.82

 Paras

-0.78

 Dalai

-0.72

BILITIES

-0.67

 interrog

-0.67

 propag

-0.67

 Recep

-0.65

 ransom

-0.65

 Reloaded

-0.64

äºº

-0.64

POSITIVE LOGITS

esville

1.17

erville

1.17

mington

1.16

mingham

1.05

heny

1.04

neapolis

1.04

ville

1.02

brook

1.01

lington

1.00

town

1.00

Activations Density 0.335%

dates, names, and locations typically seen in news articles

geographical locations and proper nouns

No Comments

No Known Activations