INDEX

Explanations

phrases related to drinking

oai_token-act-pair · gpt-3.5-turbo

references to drinking alcohol

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

CLASSIFIED

-0.78

rious

-0.69

aeda

-0.64

lat

-0.64

 accompl

-0.63

Requ

-0.63

gre

-0.63

RIC

-0.61

 suites

-0.60

 Lans

-0.60

POSITIVE LOGITS

emouth

0.84

ewater

0.74

umen

0.71

animal

0.70

çĶ°

0.70

ecast

0.69

)(

0.66

emaker

0.65

bernatorial

0.64

 Devi

0.63

Activations Density 0.000%

phrases related to drinking

references to drinking alcohol

No Comments

No Known Activations