INDEX

Explanations

questions with 'what' as the first word

oai_token-act-pair · gpt-3.5-turbo

questions that seek clarification or further information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

aku

-0.78

ishable

-0.76

 coil

-0.71

 padded

-0.69

rek

-0.69

onde

-0.68

aper

-0.68

 binge

-0.66

kee

-0.65

 worm

-0.65

POSITIVE LOGITS

 Well

1.02

 Surely

0.99

 Probably

0.97

 Nope

0.96

?:

0.95

.?

0.93

ï¿½ï¿½ï¿½ï¿½

0.90

Why

0.90

 Turns

0.89

 Perhaps

0.87

Activations Density 0.100%

questions with 'what' as the first word

questions that seek clarification or further information

No Comments

No Known Activations