INDEX

Explanations

instructions or calls to action

oai_token-act-pair · gpt-3.5-turbo

references to social media interactions and follow requests

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

 soluble

-0.72

dfx

-0.66

venge

-0.62

uable

-0.62

Exc

-0.62

 gasoline

-0.61

tu

-0.61

apon

-0.60

priority

-0.60

 condos

-0.59

POSITIVE LOGITS

 closely

0.92

 footsteps

0.81

hunt

0.73

 steps

0.72

hran

0.71

 route

0.68

Leaks

0.68

 inspires

0.65

andel

0.64

 blindly

0.64

Activations Density 0.193%

instructions or calls to action

references to social media interactions and follow requests

No Comments

No Known Activations