INDEX

Explanations

specific words and phrases related to the term "Hind", both with high and lower activations

oai_token-act-pair · gpt-3.5-turbo

references to the term "Hind" and related variants

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ptions

-0.68

 posit

-0.66

itri

-0.66

TION

-0.64

lly

-0.62

byss

-0.61

terday

-0.61

olves

-0.61

zai

-0.60

ãģĦ

-0.60

POSITIVE LOGITS

erer

1.03

rums

0.99

erers

0.96

icator

0.94

icated

0.94

elta

0.91

icates

0.87

ication

0.87

rift

0.87

aniel

0.86

Activations Density 0.081%

specific words and phrases related to the term "Hind", both with high and lower activations

references to the term "Hind" and related variants

No Comments

No Known Activations