INDEX

Explanations

positive and negative adjectives followed by ratings or numerical values

oai_token-act-pair · gpt-3.5-turbo

terms related to popularity or approval in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

but

-0.86

orie

-0.72

schild

-0.68

eatured

-0.66

ento

-0.65

regor

-0.64

estern

-0.62

ructose

-0.61

rongh

-0.60

omsky

-0.60

POSITIVE LOGITS

 nonetheless

1.65

 anyway

1.17

 anyways

1.11

 nevertheless

1.02

etheless

0.89

!.

0.84

.[

0.81

.–

0.79

 owing

0.77

Activations Density 1.118%

positive and negative adjectives followed by ratings or numerical values

terms related to popularity or approval in various contexts

No Comments

No Known Activations