INDEX

Explanations

phrases related to instructions or commands

oai_token-act-pair · gpt-3.5-turbo

specific commands or instructions related to actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 ageing

-0.72

 citiz

-0.66

 nails

-0.65

xual

-0.64

 thieves

-0.63

 aging

-0.62

 escaping

-0.61

wolves

-0.57

 diving

-0.57

 judgement

-0.55

POSITIVE LOGITS

ÂłÂłÂł

0.99

ÂłÂłÂłÂłÂłÂłÂłÂł

0.97

ÂłÂłÂłÂł

0.93

ÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂł

0.91

ÂłÂł

0.72

-.

0.70

ults

0.66

vious

0.63

îĢ

0.63

ORD

0.62

Activations Density 0.432%

phrases related to instructions or commands

specific commands or instructions related to actions

No Comments

No Known Activations