INDEX

Explanations

information related to drafting individuals in various contexts

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "drafted."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ateral

-0.75

fact

-0.74

upon

-0.71

abuse

-0.71

variable

-0.69

profit

-0.67

 rinse

-0.67

interrupted

-0.67

affer

-0.66

Brexit

-0.66

POSITIVE LOGITS

smanship

1.28

 drafted

1.08

 drafting

1.06

 drafts

1.00

 draft

0.84

Kings

0.83

ãĥ³ãĤ¸

0.78

entious

0.76

rafted

0.71

 hockey

0.71

Activations Density 0.004%

information related to drafting individuals in various contexts

occurrences of the word "drafted."

No Comments

No Known Activations