INDEX

Explanations

instances where something is "appearing to be" or "appearing to have" a certain characteristic

oai_token-act-pair · gpt-3.5-turbo

phrases expressing conditional statements or observations

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

opens

-0.74

 Rapids

-0.70

anchester

-0.68

Apps

-0.63

rowth

-0.60

ãĤ¼ãĤ¦ãĤ¹

-0.60

sterdam

-0.60

blooded

-0.60

mentation

-0.59

 winters

-0.58

POSITIVE LOGITS

 indicate

0.75

 imitate

0.72

be

0.72

ilet

0.71

 resemble

0.70

ĪĴ

0.70

 violate

0.69

 soften

0.68

 comply

0.68

 dispose

0.67

Activations Density 0.050%

instances where something is "appearing to be" or "appearing to have" a certain characteristic

phrases expressing conditional statements or observations

No Comments

No Known Activations