INDEX

Explanations

proper names, possibly focusing on individual people

oai_token-act-pair · gpt-3.5-turbo

references to specific names and key figures in the context of personal narratives or stories

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Hamb

-0.79

Java

-0.76

Ó

-0.75

åĤ

-0.75

 hamb

-0.75

hya

-0.74

 caffe

-0.74

jiang

-0.74

ãĥĺ

-0.73

MLA

-0.73

POSITIVE LOGITS

 Ross

2.35

Ross

2.11

Ros

1.85

Ros

1.85

Roz

1.47

 Rossi

1.45

ross

1.26

ROS

1.18

 Rosenthal

1.16

Cos

1.12

Activations Density 0.271%

proper names, possibly focusing on individual people

references to specific names and key figures in the context of personal narratives or stories

No Comments

No Known Activations