INDEX

Explanations

specific actions or strategies indicated by action verbs such as "moves"

oai_token-act-pair · gpt-3.5-turbo

instances of the word "moves" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Alpine

-0.68

oys

-0.66

OA

-0.66

 Gateway

-0.63

 Koran

-0.62

thia

-0.61

sson

-0.60

 Barton

-0.59

lain

-0.58

 Alaska

-0.58

POSITIVE LOGITS

peed

1.08

ivism

0.82

itters

0.80

ets

0.79

 moves

0.78

 horizont

0.77

HUD

0.76

itic

0.75

brates

0.75

hops

0.74

Activations Density 0.008%

specific actions or strategies indicated by action verbs such as "moves"

instances of the word "moves" in various contexts

No Comments

No Known Activations