INDEX

Explanations

adjectives or descriptive terms related to difficulty or severity

oai_token-act-pair · gpt-3.5-turbo

adjectives and descriptors indicating quality or condition

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ittee

-0.87

hander

-0.81

pointer

-0.79

undrum

-0.79

adder

-0.77

porter

-0.77

 tremend

-0.77

emark

-0.76

atum

-0.76

abus

-0.75

POSITIVE LOGITS

 scenarios

1.23

 versions

1.22

 environments

1.16

 situations

1.15

 setups

1.10

 alternatives

1.10

 models

1.09

 backgrounds

1.08

 solutions

1.08

 applications

1.06

Activations Density 0.518%

adjectives or descriptive terms related to difficulty or severity

adjectives and descriptors indicating quality or condition

No Comments

No Known Activations