INDEX

Explanations

descriptive phrases related to unspecified or unspecified categories, likely from structured data sources

oai_token-act-pair · gpt-3.5-turbo

phrases related to unspecified categories or general descriptors

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

senal

-0.59

'."

-0.57

.'"

-0.57

]."

-0.52

 ambush

-0.51

 MORE

-0.50

 ransom

-0.49

 damaging

-0.48

 impunity

-0.47

anwhile

-0.46

POSITIVE LOGITS

imester

0.63

 denotes

0.58

ibrary

0.57

tarian

0.57

iple

0.57

hari

0.55

 "$:/

0.54

:/

0.54

iera

0.53

iguous

0.53

Activations Density 1.320%

descriptive phrases related to unspecified or unspecified categories, likely from structured data sources

phrases related to unspecified categories or general descriptors

No Comments

No Known Activations