INDEX

Explanations

phrases related to news headlines

oai_token-act-pair · gpt-3.5-turbo

colons followed by significant statements or observations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 endeavor

-0.78

oes

-0.74

yout

-0.67

 favor

-0.67

sofar

-0.67

oku

-0.67

itol

-0.67

 realization

-0.66

Initialized

-0.65

uggest

-0.65

POSITIVE LOGITS

 Whilst

0.82

 Provided

0.81

 Firstly

0.74

 Thousands

0.74

Â£

0.74

 Â£

0.74

BBC

0.73

Scotland

0.73

 Ahead

0.69

iverpool

0.69

Activations Density 0.133%

phrases related to news headlines

colons followed by significant statements or observations

No Comments

No Known Activations