INDEX

Explanations

technical terms related to technology and programming

oai_token-act-pair · gpt-3.5-turbo

punctuation and formatting indicators in text

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

thal

-0.69

ĪĴ

-0.67

ĵĺ

-0.67

ims

-0.66

ħĭ

-0.65

ibilities

-0.65

estial

-0.63

ogun

-0.62

 sleeves

-0.62

 seams

-0.59

POSITIVE LOGITS

 ours

0.77

0.72

worth

0.71

mma

0.70

dropping

0.70

cape

0.67

giving

0.66

ablishment

0.63

calling

0.63

"+

0.63

Activations Density 0.066%

technical terms related to technology and programming

punctuation and formatting indicators in text

No Comments

No Known Activations