INDEX

Explanations

occurrences of the name "Philip" at a strong activation level

oai_token-act-pair · gpt-3.5-turbo

the name "Philip," indicating a focus on this individual's mentions throughout the text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

yrinth

-0.78

eer

-0.77

DERR

-0.74

REAM

-0.70

LOAD

-0.69

inventoryQuantity

-0.68

bnb

-0.68

Ranked

-0.66

ear

-0.66

ipeg

-0.65

POSITIVE LOGITS

 Randolph

0.90

anthrop

0.84

 Morris

0.83

osate

0.74

son

0.74

 Seymour

0.73

istine

0.71

 Philip

0.70

ophe

0.70

lectic

0.69

Activations Density 0.005%

occurrences of the name "Philip" at a strong activation level

the name "Philip," indicating a focus on this individual's mentions throughout the text

No Comments

No Known Activations