INDEX

Explanations

words related to actions or instructions

oai_token-act-pair · gpt-3.5-turbo

the phrase "in order to" or related constructs indicating purpose or intent

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

tur

-0.70

burgh

-0.68

Vice

-0.68

 Growth

-0.67

dden

-0.66

tek

-0.66

FTA

-0.65

Tan

-0.65

tiny

-0.63

Virgin

-0.63

POSITIVE LOGITS

lies

1.10

 order

0.92

liness

0.92

 orders

0.87

eering

0.83

rul

0.81

 ordering

0.80

ufact

0.78

etary

0.77

eous

0.75

Activations Density 0.025%

words related to actions or instructions

the phrase "in order to" or related constructs indicating purpose or intent

No Comments

No Known Activations