INDEX

Explanations

references to the word "odor."

oai_token-act-pair · gpt-3.5-turbo

mentions of odors

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

peak

-0.74

 notch

-0.66

ced

-0.61

 secrecy

-0.60

 Peaks

-0.60

 trust

-0.60

 selected

-0.59

 pasture

-0.58

Hay

-0.57

 transparency

-0.57

POSITIVE LOGITS

odor

1.62

owsky

1.06

inator

1.04

inators

1.01

ovy

0.91

opoulos

0.91

ovic

0.90

oslav

0.89

atus

0.87

olph

0.86

Activations Density 0.008%

references to the word "odor."

mentions of odors

No Comments

No Known Activations