INDEX

Explanations

references to walking activities

oai_token-act-pair · gpt-3.5-turbo

phrases related to walking or pedestrian activity

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

cffff

-0.70

 Saud

-0.66

encies

-0.65

illian

-0.62

 tuna

-0.58

 lobe

-0.57

iling

-0.57

 Seym

-0.57

 Hussain

-0.57

 Sovere

-0.56

POSITIVE LOGITS

through

1.25

about

1.11

ways

1.09

way

1.01

abouts

0.94

bow

0.90

ie

0.86

erb

0.84

erc

0.83

athon

0.82

Activations Density 0.040%

references to walking activities

phrases related to walking or pedestrian activity

No Comments

No Known Activations