INDEX

Explanations

information related to history

oai_token-act-pair · gpt-3.5-turbo

references to historical events and contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

ertodd

-0.87

nery

-0.85

lain

-0.85

igans

-0.74

jit

-0.74

forcement

-0.71

Beast

-0.65

cheon

-0.64

ricular

-0.64

geon

-0.64

POSITIVE LOGITS

 significance

0.99

 preservation

0.97

 inacc

0.96

 revision

0.95

orical

0.93

 accuracy

0.92

 precedent

0.91

 inaccur

0.89

 accur

0.89

 preced

0.83

Activations Density 0.058%

information related to history

references to historical events and contexts

No Comments

No Known Activations