INDEX

Explanations

references to conflicts or confrontations

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "clash" and its variations, indicating conflict or disagreement

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

mental

-1.01

©¶æ

-0.90

usage

-0.85

employment

-0.83

shown

-0.78

ufact

-0.77

incarn

-0.77

sembly

-0.74

agic

-0.74

pir

-0.74

POSITIVE LOGITS

 clash

1.30

 clashes

1.01

 Clash

0.87

halla

0.86

 looms

0.84

 showdown

0.81

 clashed

0.77

 conflicts

0.75

 brawl

0.75

 Amon

0.72

Activations Density 0.007%

references to conflicts or confrontations

occurrences of the word "clash" and its variations, indicating conflict or disagreement

No Comments

No Known Activations