INDEX

Explanations

positive adjectives

oai_token-act-pair · gpt-3.5-turbo

expressions of positive evaluations or praise

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

eters

-0.81

ople

-0.80

iper

-0.73

eter

-0.73

istan

-0.72

pper

-0.71

hyde

-0.70

hip

-0.70

hod

-0.70

 Pavilion

-0.69

POSITIVE LOGITS

enough

1.35

 enough

1.09

reads

1.02

 luck

1.00

sword

0.96

 Enough

0.92

 intentions

0.91

 Samar

0.91

ol

0.88

luck

0.86

Activations Density 0.064%

positive adjectives

expressions of positive evaluations or praise

No Comments

No Known Activations