INDEX

Explanations

references to characters and roles in storytelling or narratives

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_post_32k-oai

Configuration

jbloom/GPT2-Small-OAI-v5-32k-resid-post-SAEs/v5_32k_layer_8.pt

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

32,768

Data Type

torch.float32

Hook Name

blocks.8.hook_resid_post

Hook Layer

Architecture

standard

Context Size

Dataset

Skylion007/openwebtext

Activation Function

topk

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

PB

-0.76

iP

-0.75

 RELE

-0.72

 antiv

-0.71

Coc

-0.71

 supers

-0.70

WW

-0.70

 acron

-0.69

OCT

-0.69

 Inter

-0.69

POSITIVE LOGITS

idy

1.23

idation

1.14

ipal

1.07

opia

1.03

orks

1.03

gettable

1.03

atorial

1.01

inar

1.01

iasis

1.00

acular

1.00

Activations Density 0.603%

references to characters and roles in storytelling or narratives

No Comments

No Known Activations