INDEX

Explanations

keywords related to intensity or importance

oai_token-act-pair · gpt-3.5-turbo

words related to visual imagery and nuclear topics

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

çĦ

-0.70

slot

-0.65

OTOS

-0.63

ILL

-0.61

 clipboard

-0.61

gone

-0.61

skirts

-0.59

 McMaster

-0.58

Discover

-0.57

 Clarks

-0.56

POSITIVE LOGITS

ivid

1.34

itri

0.98

escent

0.93

terness

0.91

 agre

0.87

itives

0.83

itionally

0.82

atives

0.81

uation

0.81

gments

0.79

Activations Density 0.009%

keywords related to intensity or importance

words related to visual imagery and nuclear topics

No Comments

No Known Activations