INDEX

Explanations

mentions of combinations of letters or words that form a specific sequence

oai_token-act-pair · gpt-3.5-turbo

terms related to tetrahedra and deducible models

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

FUL

-0.82

Ø©

-0.74

 Dragonbound

-0.64

istas

-0.63

²¾

-0.62

eland

-0.62

IER

-0.61

ATURES

-0.61

ERY

-0.60

ãĥ¥

-0.60

POSITIVE LOGITS

icated

0.90

ication

0.85

icate

0.84

uctive

0.82

ria

0.81

icating

0.78

icates

0.77

uction

0.77

raph

0.76

itable

0.76

Activations Density 0.117%

mentions of combinations of letters or words that form a specific sequence

terms related to tetrahedra and deducible models

No Comments

No Known Activations