INDEX

Explanations

addresses or locations described with block numbers

oai_token-act-pair · gpt-3.5-turbo

references to specific locations or blocks in a city

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

emia

-0.66

eer

-0.61

 diction

-0.60

 programmes

-0.60

Versions

-0.59

 submar

-0.58

 Countries

-0.58

 coward

-0.58

 fatigue

-0.58

 bureaucr

-0.57

POSITIVE LOGITS

 radius

0.90

tyard

0.90

 southwest

0.88

 northwest

0.85

heast

0.83

 east

0.79

yards

0.77

 bounded

0.76

 northeast

0.75

north

0.75

Activations Density 0.032%

addresses or locations described with block numbers

references to specific locations or blocks in a city

No Comments

No Known Activations