INDEX

Explanations

locations or names related to places

oai_token-act-pair · gpt-3.5-turbo

references to the name "Wood" in various contexts

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

uates

-0.74

ional

-0.72

ysis

-0.70

 aver

-0.67

 disguise

-0.65

 analges

-0.64

itia

-0.64

 disenfranch

-0.63

uate

-0.63

 anonymity

-0.62

POSITIVE LOGITS

ruff

1.54

bridge

1.24

stock

1.22

working

1.18

bury

1.13

lands

1.13

leigh

1.12

cutting

1.11

burn

1.11

row

1.11

Activations Density 0.030%

locations or names related to places

references to the name "Wood" in various contexts

No Comments

No Known Activations