INDEX

Explanations

numerical values with associated units

oai_token-act-pair · gpt-3.5-turbo

specific symbols and terms related to logic notation or programming syntax

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

ritic

-0.79

ahime

-0.78

abies

-0.78

ament

-0.77

sky

-0.74

andro

-0.74

rule

-0.73

idy

-0.71

eri

-0.70

owl

-0.70

POSITIVE LOGITS

âĢ¢âĢ¢âĢ¢âĢ¢

0.97

âĢ¢âĢ¢

0.88

Ħ¢

0.78

 ÃĹ

0.73

Ģ

0.71

++++++++++++++++

0.70

Â·Â·

0.70

 array

0.63

 lact

0.63

 layers

0.61

Activations Density 0.010%

numerical values with associated units

specific symbols and terms related to logic notation or programming syntax

No Comments

No Known Activations