INDEX

Explanations

words related to negation, uncertainty, or questioning

oai_token-act-pair · gpt-3.5-turbo

the negative contraction "don't."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 Dickinson

-0.75

 Compass

-0.65

IAS

-0.64

 Reborn

-0.60

 Maiden

-0.60

 Starts

-0.58

states

-0.58

tips

-0.58

ISON

-0.57

 Pike

-0.56

POSITIVE LOGITS

 bother

1.04

 necessarily

1.03

 deserve

0.99

 discriminate

0.99

 hesitate

0.98

Â¨

0.98

Í

0.98

 care

0.96

 intend

0.96

 CARE

0.96

Activations Density 0.098%

words related to negation, uncertainty, or questioning

the negative contraction "don't."

No Comments

No Known Activations