INDEX

Explanations

numerical data or figures

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 6-res_post_32k-oai

Configuration

jbloom/GPT2-Small-OAI-v5-32k-resid-post-SAEs/v5_32k_layer_6.pt

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

32,768

Data Type

torch.float32

Hook Name

blocks.6.hook_resid_post

Hook Layer

Architecture

standard

Context Size

Dataset

Skylion007/openwebtext

Activation Function

topk

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

iques

-1.00

ique

-0.96

ickers

-0.90

istics

-0.87

ettes

-0.81

ezvous

-0.80

ivia

-0.78

izers

-0.78

nect

-0.78

alyst

-0.76

POSITIVE LOGITS

lda

1.16

anwhile

0.77

alloc

0.72

upiter

0.70

rals

0.70

lems

0.70

hovah

0.70

ppelin

0.70

 fixme

0.69

BLIC

0.67

Activations Density 0.088%

numerical data or figures

No Comments

No Known Activations