INDEX

Explanations

phrases related to providing important information or insights

oai_token-act-pair · gpt-3.5-turbo

phrases indicating a portion or a small piece of a larger issue

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 Nationwide

-0.74

ufact

-0.72

Zed

-0.67

ruciating

-0.65

 Commodore

-0.65

ively

-0.64

Bom

-0.64

effic

-0.63

imposed

-0.63

 Recall

-0.62

POSITIVE LOGITS

jar

1.15

 toes

1.08

 iceberg

1.01

tip

1.00

tip

0.97

ster

0.94

sy

0.92

sters

0.91

toe

0.90

ariat

0.83

Activations Density 0.030%

phrases related to providing important information or insights

phrases indicating a portion or a small piece of a larger issue

No Comments

No Known Activations