INDEX

Explanations

text related to units of measurement

oai_token-act-pair · gpt-3.5-turbo

numerical representations often related to measurements or quantities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 notices

-0.82

 tender

-0.76

 sooner

-0.75

 poaching

-0.74

nod

-0.73

 nods

-0.72

 amounts

-0.70

 Graves

-0.69

 markings

-0.69

 Reconstruction

-0.69

POSITIVE LOGITS

etc

1.18

AU

1.16

each

1.13

Flickr

1.09

Â£

1.06

hr

1.06

sq

1.04

EVA

1.04

month

1.02

Activations Density 0.032%

text related to units of measurement

numerical representations often related to measurements or quantities

No Comments

No Known Activations