INDEX

Explanations

the name "Don" with different emphasis scores

oai_token-act-pair · gpt-3.5-turbo

the word "don" 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 @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Flight

-0.66

LCS

-0.65

EStream

-0.65

ULT

-0.64

TIT

-0.63

 Healing

-0.61

 NIGHT

-0.61

 Nightmare

-0.60

 Instruction

-0.60

 Claw

-0.60

POSITIVE LOGITS

don

1.06

nell

1.05

nie

1.03

nel

1.01

etheless

0.99

neau

0.97

kie

0.97

ates

0.95

stant

0.94

ctory

0.94

Activations Density 0.007%

the name "Don" with different emphasis scores

the word "don" in various contexts

No Comments

No Known Activations