INDEX

Explanations

phrases related to conjunctions, particularly the word 'and'

oai_token-act-pair · gpt-3.5-turbo

conjunctions that indicate continuity or connection between ideas

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

ULTS

-0.81

ODY

-0.75

ESE

-0.75

verts

-0.73

runs

-0.71

 Appearance

-0.71

 Lines

-0.70

ults

-0.69

isance

-0.69

 Levels

-0.69

POSITIVE LOGITS

rogen

1.12

 costly

1.09

 lucrative

1.06

 controversial

1.06

 sometimes

1.05

oft

1.04

 unpredictable

1.04

 potentially

1.03

 expensive

1.02

 possibly

1.01

Activations Density 0.109%

phrases related to conjunctions, particularly the word 'and'

conjunctions that indicate continuity or connection between ideas

No Comments

No Known Activations