INDEX

Explanations

phrases or sections marked by a sequence of repeated characters or symbols

oai_token-act-pair · gpt-3.5-turbo

decorative patterns or symbols

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

amorph

-0.71

 divers

-0.65

 cycling

-0.65

cape

-0.65

mates

-0.64

 subord

-0.64

 leading

-0.61

 Bridgewater

-0.60

 graft

-0.60

 mete

-0.60

POSITIVE LOGITS

_______

1.54

______

1.52

____

1.45

********

1.44

_____

1.43

************

1.39

********************************

1.38

___

1.35

****************

1.29

________

1.29

Activations Density 0.032%

phrases or sections marked by a sequence of repeated characters or symbols

decorative patterns or symbols

No Comments

No Known Activations