INDEX

Explanations

political terms and institutions

oai_token-act-pair · gpt-3.5-turbo

references to political entities and structures

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

lopp

-0.63

 Trop

-0.58

GW

-0.55

 Bree

-0.54

 highs

-0.54

 entreprene

-0.52

 prototypes

-0.52

Lov

-0.51

IPO

-0.50

struct

-0.50

POSITIVE LOGITS

 whom

0.90

who

0.85

 alike

0.79

 respectively

0.72

 irrespective

0.72

who

0.71

edom

0.70

ãĢĤ

0.70

0.68

 subscribing

0.66

Activations Density 0.447%

political terms and institutions

references to political entities and structures

No Comments

No Known Activations