INDEX

Explanations

phrases related to predictable patterns or routines

oai_token-act-pair · gpt-3.5-turbo

phrases indicating routine or common occurrences

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

lez

-0.84

sten

-0.76

mented

-0.75

Stars

-0.74

reth

-0.71

jection

-0.70

mentation

-0.70

ASED

-0.69

asus

-0.68

onics

-0.68

POSITIVE LOGITS

 suspects

0.93

 disclaim

0.90

 disclaimer

0.87

 caveats

0.84

 deviations

0.81

ITIES

0.73

 assortment

0.73

 deviation

0.70

tenance

0.70

ãĥķãĤ©

0.69

Activations Density 0.021%

phrases related to predictable patterns or routines

phrases indicating routine or common occurrences

No Comments

No Known Activations