INDEX

Explanations

specific symbols or non-standard characters used as markers or separators

oai_token-act-pair · gpt-3.5-turbo

indications of links or references, particularly in a structured format

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

edo

-0.79

aque

-0.73

 dispers

-0.72

 Bengal

-0.72

xus

-0.66

eki

-0.65

ured

-0.64

urable

-0.64

 popul

-0.64

 Antar

-0.64

POSITIVE LOGITS

<<

0.91

lations

0.88

[[

0.86

SOURCE

0.86

Â¢

0.80

>>>>>>>>

0.80

>>

0.79

PER

0.78

PIN

0.77

HEAD

0.77

Activations Density 0.010%

specific symbols or non-standard characters used as markers or separators

indications of links or references, particularly in a structured format

No Comments

No Known Activations