INDEX

Explanations

expressions stating or emphasizing a main idea or argument

oai_token-act-pair · gpt-3.5-turbo

references to central themes or arguments in a discussion

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

NetMessage

-0.76

uthor

-0.74

ctic

-0.72

 Seym

-0.71

reditary

-0.70

 Carbuncle

-0.69

ndum

-0.66

eco

-0.65

ãĥ³ãĤ¸

-0.65

ESE

-0.65

POSITIVE LOGITS

iasis

0.93

 Reyes

0.85

lessness

0.83

lessly

0.80

points

0.78

posts

0.76

point

0.76

ifications

0.72

otics

0.70

manship

0.67

Activations Density 0.032%

expressions stating or emphasizing a main idea or argument

references to central themes or arguments in a discussion

No Comments

No Known Activations