INDEX

Explanations

phrases related to patterns or sequences

oai_token-act-pair · gpt-3.5-turbo

patterns related to data analysis and recognition

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

ournal

-0.74

©¶æ¥µ

-0.70

abase

-0.67

tein

-0.63

 grate

-0.62

ppard

-0.62

anamo

-0.62

 cultured

-0.62

ktop

-0.61

enhagen

-0.61

POSITIVE LOGITS

 guiActiveUnfocused

0.95

gradient

0.74

natureconservancy

0.73

angular

0.71

 unfold

0.70

 convergence

0.69

 converge

0.67

ciation

0.67

 Yards

0.66

 ACTIONS

0.65

Activations Density 1.010%

phrases related to patterns or sequences

patterns related to data analysis and recognition

No Comments

No Known Activations