INDEX

Explanations

phrases related to technical and structured topics, possibly in a list format

oai_token-act-pair · gpt-3.5-turbo

structures related to various topics and concepts

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

hement

-1.12

uly

-0.72

boro

-0.65

abouts

-0.64

ennes

-0.64

dn

-0.62

lected

-0.62

etta

-0.62

acan

-0.62

ilipp

-0.62

POSITIVE LOGITS

ÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂł

0.95

ÂłÂłÂłÂłÂłÂłÂłÂł

0.82

ÂłÂłÂł

0.75

ccording

0.69

ÂłÂłÂłÂł

0.63

ÂłÂł

0.63

ãĤ¨ãĥ«

0.61

 Edit

0.60

 marrow

0.59

 Appearance

0.59

Activations Density 0.215%

phrases related to technical and structured topics, possibly in a list format

structures related to various topics and concepts

No Comments

No Known Activations