INDEX

Explanations

codes or identification numbers in a particular format

oai_token-act-pair · gpt-3.5-turbo

sequences of numbers and characters, likely related to codes or identifiers

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

DonaldTrump

-0.80

worldly

-0.68

 urgently

-0.67

Ü

-0.66

 globalization

-0.62

 compan

-0.61

 multinational

-0.59

 sausage

-0.59

 weekday

-0.58

 carefully

-0.57

POSITIVE LOGITS

fy

0.77

 respectively

0.76

ea

0.73

phrine

0.73

fu

0.72

0.70

0.69

}:

0.69

0.68

Activations Density 0.027%

codes or identification numbers in a particular format

sequences of numbers and characters, likely related to codes or identifiers

No Comments

No Known Activations