INDEX

Explanations

references to specific titles or names

oai_token-act-pair · gpt-3.5-turbo

references to titles and descriptions of creative works

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

arching

-0.79

conservancy

-0.72

successfully

-0.70

packing

-0.66

bending

-0.66

icer

-0.63

ittal

-0.63

ngth

-0.60

aults

-0.59

fair

-0.59

POSITIVE LOGITS

=""

0.84

sic

0.81

Ô

0.78

Â¯

0.76

="

0.76

Said

0.70

 Dictionary

0.70

lua

0.69

Quantity

0.69

èĥ

0.68

Activations Density 0.364%

references to specific titles or names

references to titles and descriptions of creative works

No Comments

No Known Activations