INDEX

Explanations

references to the word "mold" in the context of shaping or forming

oai_token-act-pair · gpt-3.5-turbo

references to mold

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

 veter

-0.77

ixel

-0.65

 Kamp

-0.64

 Kurds

-0.64

 Franks

-0.59

FLAG

-0.59

Es

-0.58

 Kurd

-0.58

 respir

-0.58

illation

-0.58

POSITIVE LOGITS

ovan

0.78

ificial

0.77

estial

0.73

eful

0.72

ctuary

0.72

elected

0.68

omorphic

0.68

stant

0.67

 iceberg

0.67

erest

0.66

Activations Density 0.000%

references to the word "mold" in the context of shaping or forming

references to mold

No Comments

No Known Activations