INDEX

Explanations

the word "UR."

oai_token-act-pair · gpt-3.5-turbo

repeated instances of the term "UR."

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

 Schultz

-0.74

Eb

-0.71

 BaseType

-0.70

 Schn

-0.69

etts

-0.67

 Sands

-0.64

olean

-0.64

 Kissinger

-0.63

 Ghana

-0.61

hed

-0.61

POSITIVE LOGITS

UR

1.14

POSE

1.05

 confir

1.04

GER

0.99

BLE

0.97

ARCH

0.92

pees

0.91

RR

0.90

DAY

0.88

pee

0.88

Activations Density 0.009%

the word "UR."

repeated instances of the term "UR."

No Comments

No Known Activations