INDEX

Explanations

phrases related to actions involving a change in structure or status

oai_token-act-pair · gpt-3.5-turbo

instances of the prefix "de-" indicating a process of reversal or negation

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

LW

-0.90

 Dickinson

-0.75

Ys

-0.66

 Dillon

-0.62

uez

-0.62

 Thick

-0.61

ANGE

-0.61

 galleries

-0.61

 SHARES

-0.61

 unto

-0.61

POSITIVE LOGITS

fact

1.22

emphasis

1.19

escal

1.18

cert

1.00

capt

0.99

vacc

0.97

funding

0.97

register

0.96

branded

0.95

radical

0.95

Activations Density 0.035%

phrases related to actions involving a change in structure or status

instances of the prefix "de-" indicating a process of reversal or negation

No Comments

No Known Activations