INDEX

Explanations

exclamations or fillers used in speech

oai_token-act-pair · gpt-3.5-turbo

frequent filler words and brand names

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

Kings

-0.71

rawn

-0.67

 Kings

-0.65

mented

-0.63

 glossy

-0.63

Mel

-0.63

Lex

-0.63

 Papers

-0.62

ASED

-0.62

 Chennai

-0.62

POSITIVE LOGITS

Uh

3.08

Uh

2.02

Hut

1.85

uh

1.81

arge

1.76

Um

1.46

Um

1.40

 standpoint

1.05

 Hutch

1.01

Naw

1.00

Activations Density 0.049%

exclamations or fillers used in speech

frequent filler words and brand names

No Comments

No Known Activations