INDEX

Explanations

phrases or terms indicating no change or lack of change

oai_token-act-pair · gpt-3.5-turbo

phrases indicating a lack of change or consistency

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

ardi

-0.69

Gate

-0.66

lan

-0.63

ast

-0.58

weather

-0.58

alla

-0.58

 Muse

-0.57

gee

-0.57

 Syndrome

-0.57

io

-0.56

POSITIVE LOGITS

 unchanged

3.70

 unaffected

2.12

 untouched

2.05

 intact

1.70

 unch

1.26

 identical

1.18

 unim

1.17

 Unch

1.13

 uninterrupted

1.13

 unrem

1.12

Activations Density 0.016%

phrases or terms indicating no change or lack of change

phrases indicating a lack of change or consistency

No Comments

No Known Activations