INDEX

Explanations

terms related to government actions or entities

oai_token-act-pair · gpt-3.5-turbo

frequent references to the government and its actions

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Junction

-0.79

Rank

-0.73

catentry

-0.73

sbm

-0.71

hots

-0.71

Strange

-0.71

 Heaven

-0.69

luaj

-0.68

 Daylight

-0.68

Nig

-0.68

POSITIVE LOGITS

 agencies

0.94

 bureaucracy

0.94

 coffers

0.90

 contractor

0.90

 shutdown

0.89

 Accountability

0.88

 agency

0.84

 watchdog

0.84

 treasury

0.80

 government

0.79

Activations Density 0.125%

terms related to government actions or entities

frequent references to the government and its actions

No Comments

No Known Activations