INDEX

Explanations

programming code elements

oai_token-act-pair · gpt-3.5-turbo

symbols or characters commonly used in programming or code syntax

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

arna

-0.81

 eleph

-0.72

 ende

-0.69

 pronoun

-0.66

amia

-0.65

oun

-0.64

 disbanded

-0.64

 consum

-0.62

oran

-0.62

 rooting

-0.61

POSITIVE LOGITS

Pg

1.02

tm

0.88

mop

0.83

bh

0.81

-+

0.79

sup

0.77

ouls

0.76

Temperature

0.76

rt

0.76

Activations Density 0.049%

programming code elements

symbols or characters commonly used in programming or code syntax

No Comments

No Known Activations