INDEX

Explanations

words related to being hired or working for someone

oai_token-act-pair · gpt-3.5-turbo

references to employment

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 replay

-0.69

aline

-0.68

 Replay

-0.68

Heart

-0.62

hide

-0.61

 VIDEOS

-0.61

 Compatibility

-0.61

 kickoff

-0.59

 goodbye

-0.58

blu

-0.58

POSITIVE LOGITS

 employed

3.55

 employ

2.13

 employs

1.95

employed

1.94

 employing

1.92

 utilized

1.81

 hired

1.57

 Employ

1.50

 deployed

1.42

Employ

1.40

Activations Density 0.015%

words related to being hired or working for someone

references to employment

No Comments

No Known Activations