INDEX

Explanations

expressions related to successful outcomes or achievements

oai_token-act-pair · gpt-3.5-turbo

expressions related to achieving success or hitting a target

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

iterator

-0.74

gio

-0.73

tymology

-0.65

wcsstore

-0.65

lance

-0.65

href

-0.63

urdue

-0.62

shown

-0.61

lege

-0.60

Vo

-0.60

POSITIVE LOGITS

 stride

1.07

 snag

1.00

 nail

0.90

 pavement

0.84

 iceberg

0.84

 hardest

0.82

 brakes

0.82

 bounce

0.77

 itch

0.76

 button

0.76

Activations Density 0.142%

expressions related to successful outcomes or achievements

expressions related to achieving success or hitting a target

No Comments

No Known Activations