INDEX

Explanations

text related to investigations or examining matters

oai_token-act-pair · gpt-3.5-turbo

instances of the word "the."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

peak

-0.83

cil

-0.75

elected

-0.71

LEASE

-0.68

ranged

-0.65

loaded

-0.65

hops

-0.65

--+

-0.65

[|

-0.64

upe

-0.64

POSITIVE LOGITS

 possibility

1.49

 feasibility

1.47

 legality

1.43

 matter

1.41

 merits

1.40

 propri

1.30

 implications

1.26

 ramifications

1.26

 validity

1.24

 circumstances

1.21

Activations Density 0.185%

text related to investigations or examining matters

instances of the word "the."

No Comments

No Known Activations