INDEX

Explanations

words related to mistakes or incorrect actions

oai_token-act-pair · gpt-3.5-turbo

words and concepts related to clothing and potential misunderstandings

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 Atomic

-0.73

 pian

-0.69

 simulac

-0.66

Lum

-0.64

ano

-0.64

 Crimes

-0.63

 Assass

-0.62

 renown

-0.61

 inept

-0.61

Nos

-0.60

POSITIVE LOGITS

dress

3.09

strap

1.70

flow

1.35

mis

1.33

enance

1.23

gear

1.22

nings

1.13

flows

1.09

plate

1.08

cloth

1.05

Activations Density 0.055%

words related to mistakes or incorrect actions

words and concepts related to clothing and potential misunderstandings

No Comments

No Known Activations