INDEX

Explanations

words related to criticism or evaluation, particularly focusing on negative aspects

oai_token-act-pair · gpt-3.5-turbo

repeated use of the conjunction "and" in various contexts

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

ĸļ

-0.82

edia

-0.77

 Wichita

-0.71

wolf

-0.71

ETS

-0.69

Newsletter

-0.68

 Lucius

-0.68

oused

-0.68

oker

-0.68

ainted

-0.67

POSITIVE LOGITS

 configurations

0.97

 combinations

0.92

 acron

0.89

 accomplishments

0.88

 accol

0.88

 perm

0.88

idi

0.83

 inventions

0.83

 thous

0.83

 innovations

0.83

Activations Density 0.289%

words related to criticism or evaluation, particularly focusing on negative aspects

repeated use of the conjunction "and" in various contexts

No Comments

No Known Activations