INDEX

Explanations

symbols and characters in a specific format commonly used in technical or encoded text

oai_token-act-pair · gpt-3.5-turbo

occurrences of the registered trademark symbol

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

pool

-0.70

 dispers

-0.68

 downgrade

-0.67

 description

-0.66

 report

-0.66

 pool

-0.65

 Pass

-0.63

 roll

-0.62

 crocod

-0.61

 Pool

-0.60

POSITIVE LOGITS

®

4.50

¯

2.17

Ń

2.01

¬

1.90

İ

1.86

«

1.79

·

1.79

ħ

1.79

ľ

1.71

ª

1.69

Activations Density 0.011%

symbols and characters in a specific format commonly used in technical or encoded text

occurrences of the registered trademark symbol

No Comments

No Known Activations