INDEX

Explanations

technical commands or instructions

oai_token-act-pair · gpt-3.5-turbo

code-related commands and flags

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

pak

-0.64

 apples

-0.63

qus

-0.59

emet

-0.59

ink

-0.58

 contag

-0.57

 Whatever

-0.57

 stitch

-0.56

 explan

-0.55

 grapes

-0.54

POSITIVE LOGITS

/-

0.91

=~=~

0.80

/+

0.78

¥µ

0.75

ħĭ

0.74

FactoryReloaded

0.69

VL

0.67

ãĥĭ

0.66

yright

0.66

BuyableInstoreAndOnline

0.66

Activations Density 0.116%

technical commands or instructions

code-related commands and flags

No Comments

No Known Activations