INDEX

Explanations

phrases related to news headlines or article titles

oai_token-act-pair · gpt-3.5-turbo

phrases related to continuation or progression in various contexts

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

 superf

-0.74

 dipping

-0.71

 arrivals

-0.71

 prohib

-0.71

 affected

-0.71

 erase

-0.71

 pointed

-0.71

 sealed

-0.70

 masked

-0.69

 inflated

-0.69

POSITIVE LOGITS

selves

1.13

 Them

1.08

itialized

0.99

Us

0.98

 Each

0.97

Its

0.95

 Than

0.95

slaught

0.95

 Their

0.93

 Course

0.91

Activations Density 0.164%

phrases related to news headlines or article titles

phrases related to continuation or progression in various contexts

No Comments

No Known Activations