INDEX

Explanations

unrelated or adjacent items or incidents

oai_token-act-pair · gpt-3.5-turbo

references to items or events that are not connected or relevant to the main topic

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

ilt

-0.67

Â®

-0.62

hol

-0.62

 Press

-0.60

 patience

-0.59

 PRESS

-0.59

iership

-0.58

 Reef

-0.58

estro

-0.57

iest

-0.57

POSITIVE LOGITS

 unrelated

3.61

 related

1.54

 irrelevant

1.34

 incompatible

1.29

 innocuous

1.25

 identical

1.23

 unaffected

1.23

 unspecified

1.21

 unexplained

1.17

 incidental

1.15

Activations Density 0.012%

unrelated or adjacent items or incidents

references to items or events that are not connected or relevant to the main topic

No Comments

No Known Activations