INDEX

Explanations

names or terms related to specific locations or cultures

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res_post_32k-oai

Configuration

jbloom/GPT2-Small-OAI-v5-32k-resid-post-SAEs/v5_32k_layer_10.pt

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

32,768

Data Type

torch.float32

Hook Name

blocks.10.hook_resid_post

Hook Layer

Architecture

standard

Context Size

Dataset

Skylion007/openwebtext

Activation Function

topk

Embeds

IFrame

Link

Negative Logits

BILITY

-0.76

enance

-0.68

 Jarrett

-0.64

orsi

-0.64

 lett

-0.63

inally

-0.61

 Astro

-0.60

romeda

-0.59

 Compass

-0.57

offer

-0.55

POSITIVE LOGITS

ny

1.14

cess

1.07

osaurs

1.04

ners

1.04

jiang

1.01

amic

0.99

atown

0.99

kee

0.96

ning

0.93

ghai

0.91

Activations Density 0.357%

names or terms related to specific locations or cultures

No Known Activations