INDEX

Explanations

phrases related to entities or actions being in the middle or center

oai_token-act-pair · gpt-3.5-turbo

phrases that contain variations of "the" and different numeric references, indicating a focus on structured occurrences or patterns

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

avan

-0.65

PsyNetMessage

-0.64

nesday

-0.63

iott

-0.63

orsi

-0.62

Ç

-0.62

ingly

-0.61

EH

-0.60

ulner

-0.57

mania

-0.56

POSITIVE LOGITS

 trenches

1.07

 headlights

0.96

 womb

0.94

 wilderness

0.90

 midst

0.90

 closet

0.88

 woods

0.87

 saddle

0.85

 pudding

0.84

sun

0.82

Activations Density 0.095%

phrases related to entities or actions being in the middle or center

phrases that contain variations of "the" and different numeric references, indicating a focus on structured occurrences or patterns

No Comments

No Known Activations