INDEX

Explanations

words related to specific names or labels

oai_token-act-pair · gpt-3.5-turbo

letters and specific non-word characters

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

-0.68

£ı

-0.64

CLASSIFIED

-0.63

 plague

-0.63

 Cros

-0.62

ctors

-0.61

@@@@

-0.59

GROUND

-0.59

ONSORED

-0.59

IRE

-0.57

POSITIVE LOGITS

antes

0.90

ifying

0.84

ida

0.82

iculture

0.79

ijn

0.79

adel

0.79

aceous

0.79

seys

0.77

illo

0.77

erton

0.77

Activations Density 0.129%

words related to specific names or labels

letters and specific non-word characters

No Comments

No Known Activations