INDEX

Explanations

words related to long-range objects or activities

oai_token-act-pair · gpt-3.5-turbo

references to long-range missiles

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

advertisement

-0.76

roma

-0.72

 behold

-0.68

 root

-0.63

Hyper

-0.61

ODUCT

-0.59

 glor

-0.59

older

-0.58

Fe

-0.58

 pronounced

-0.56

POSITIVE LOGITS

range

1.27

finder

0.97

ollah

0.89

xual

0.86

 Rover

0.81

tenance

0.80

anges

0.80

erness

0.80

 fortun

0.77

velength

0.77

Activations Density 0.014%

words related to long-range objects or activities

references to long-range missiles

No Comments

No Known Activations