INDEX

Explanations

words related to expressing strong opinions or recommendations

oai_token-act-pair · gpt-3.5-turbo

expressions of strong opinion or condemnation

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

 wonderfully

-0.73

emetery

-0.69

 Olymp

-0.69

 unfocusedRange

-0.69

Reviewer

-0.68

 tragically

-0.66

 magically

-0.66

 abundantly

-0.66

ammy

-0.65

 Parenthood

-0.64

POSITIVE LOGITS

 disagreed

1.12

 objected

1.04

 disagrees

1.03

 disagree

1.02

 opposes

1.02

 opposed

1.02

 oppose

1.01

 resisted

1.01

 denies

1.00

 disapprove

0.98

Activations Density 0.067%

words related to expressing strong opinions or recommendations

expressions of strong opinion or condemnation

No Comments

No Known Activations