INDEX

Explanations

phrases related to a particular country or place

oai_token-act-pair · gpt-3.5-turbo

the presence of the term "va" which likely relates to a specific identifier or classification within a context (possibly referring to variables or entities in a system)

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

cases

-0.79

papers

-0.77

dress

-0.76

brow

-0.71

taker

-0.71

sworth

-0.70

wolves

-0.70

lins

-0.70

liners

-0.69

sticks

-0.69

POSITIVE LOGITS

 Scotia

0.97

Äį

0.95

ignt

0.89

uthor

0.89

 Hels

0.88

qua

0.87

ille

0.87

quez

0.87

ices

0.79

apor

0.79

Activations Density 0.010%

phrases related to a particular country or place

the presence of the term "va" which likely relates to a specific identifier or classification within a context (possibly referring to variables or entities in a system)

No Comments

No Known Activations