INDEX

Explanations

action verbs in past tense with '-ed' suffix

oai_token-act-pair · gpt-3.5-turbo

negative statements or assessments related to social issues

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

 Mississ

-0.73

 Thrones

-0.73

 Canaver

-0.73

 Metall

-0.71

 Stones

-0.70

 Schwar

-0.69

 Rolls

-0.69

 Isle

-0.69

 Skywalker

-0.68

 Corps

-0.67

POSITIVE LOGITS

terday

1.26

theless

1.24

anwhile

1.10

etheless

1.09

maxwell

0.89

 guiActiveUn

0.80

bsite

0.80

mosp

0.78

ividual

0.77

ickr

0.77

Activations Density 0.206%

action verbs in past tense with '-ed' suffix

negative statements or assessments related to social issues

No Comments

No Known Activations