You can deploy this model and start earning money today!

rufimelo/
vulnerable_code_qwen_coder_standard_16384_10M

Name: vulnerable_code_qwen_coder_standard_16384_10M
Availability: InStock

sae_lens

sparse-autoencoder

mechanistic-interpretability

sae

Views

No views yet

Sparse Autoencoders for Qwen/Qwen2.5-7B-Instruct

This repository contains 1 Sparse Autoencoder(s) (SAE) trained using SAELens.

Model Details

Property	Value
Base Model	`Qwen/Qwen2.5-7B-Instruct`
Architecture	`standard`
Input Dimension	3584
SAE Dimension	16384
Training Dataset	`TQRG/DeltaSecommits_qwen-2.5-7b-instruct_tokenized_v2_vulnerable`

Available Hook Points

Hook Point
`blocks.11.hook_resid_post`

Usage

python

from sae_lens import SAE

# Load an SAE for a specific hook point
sae, cfg_dict, sparsity = SAE.from_pretrained(
    release="rufimelo/vulnerable_code_qwen_coder_standard_16384_10M",
    sae_id="blocks.11.hook_resid_post"  # Choose from available hook points above
)

# Use with TransformerLens
from transformer_lens import HookedTransformer

model = HookedTransformer.from_pretrained("Qwen/Qwen2.5-7B-Instruct")

# Get activations and encode
_, cache = model.run_with_cache("your text here")
activations = cache["blocks.11.hook_resid_post"]
features = sae.encode(activations)

Files

blocks.11.hook_resid_post/cfg.json - SAE configuration
blocks.11.hook_resid_post/sae_weights.safetensors - Model weights
blocks.11.hook_resid_post/sparsity.safetensors - Feature sparsity statistics

Training

These SAEs were trained with SAELens version 6.26.2.