INDEX

Explanations

quotations

oai_token-act-pair · gpt-3.5-turbo

instances of quoted speech and attributions

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

anging

-0.73

angs

-0.66

aukee

-0.61

atown

-0.60

rafted

-0.59

 tidal

-0.59

ravel

-0.58

 lighthouse

-0.58

usercontent

-0.58

icum

-0.57

POSITIVE LOGITS

Shape

0.73

'[

0.66

 citing

0.66

 sarcast

0.65

"#

0.64

 quotes

0.64

ITV

0.63

FSA

0.62

 bluntly

0.61

said

0.60

Activations Density 1.440%

quotations

instances of quoted speech and attributions

No Comments

No Known Activations