INDEX

Explanations

names or proper nouns

oai_token-act-pair · gpt-3.5-turbo

patterns of repeating letters or syllables in words

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

 shoulder

-0.72

 prost

-0.72

 multim

-0.71

 bureaucr

-0.71

 corridors

-0.70

 Directions

-0.69

 envelope

-0.67

 DIRECT

-0.66

âĢ¢âĢ¢

-0.65

 rift

-0.64

POSITIVE LOGITS

rex

0.99

ember

0.99

eware

0.98

atron

0.98

rill

0.98

bach

0.97

ador

0.96

itan

0.95

orah

0.93

artz

0.93

Activations Density 0.163%

names or proper nouns

patterns of repeating letters or syllables in words

No Comments

No Known Activations