INDEX

Explanations

words related to specific locations, such as cities and provinces in Eastern Europe

oai_token-act-pair · gpt-3.5-turbo

mentions of the region Donetsk and its related terms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

xon

-0.85

HAHAHAHA

-0.73

pir

-0.72

Haunted

-0.72

cohol

-0.71

 toug

-0.70

gently

-0.70

icone

-0.68

Reloaded

-0.67

cock

-0.67

POSITIVE LOGITS

 Donetsk

0.90

ansk

0.84

arna

0.74

 Croat

0.73

ÅĤ

0.69

nance

0.68

 Croatian

0.68

 Croatia

0.67

 Graveyard

0.65

 Speedway

0.65

Activations Density 0.017%

words related to specific locations, such as cities and provinces in Eastern Europe

mentions of the region Donetsk and its related terms

No Comments

No Known Activations