INDEX

Explanations

random combinations of characters that don't seem to follow a specific pattern or meaning

oai_token-act-pair · gpt-3.5-turbo

sequences of unique characters or symbols

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

geries

-0.90

icides

-0.83

eworld

-0.78

rha

-0.77

NetMessage

-0.76

nels

-0.75

 tradem

-0.75

uld

-0.74

fight

-0.74

orget

-0.74

POSITIVE LOGITS

×

1.83

×

1.73

×ķ

1.68

×Ļ

1.62

×Ļ×

1.60

×ª

1.58

×ľ

1.57

×¨

1.51

×Ĳ

1.50

×©

1.50

Activations Density 0.005%

random combinations of characters that don't seem to follow a specific pattern or meaning

sequences of unique characters or symbols

No Comments

No Known Activations