Add links to paper, code, and project page

#7
by nielsr HF Staff - opened
Files changed (1) hide show
  1. README.md +14 -10
README.md CHANGED
@@ -1,21 +1,25 @@
1
  ---
2
- license: apache-2.0
3
- library_name: steerling
4
- tags:
5
- - causal-diffusion
6
- - interpretability
7
- - concept-steering
8
- - masked-diffusion
9
- - block-causal
10
  language:
11
- - en
 
 
12
  pipeline_tag: text-generation
 
 
 
 
 
 
13
  ---
14
 
15
  # Steerling-8B:
16
 
17
  **An interpretable causal diffusion language model with concept steering.**
18
 
 
 
 
 
19
  Steerling-8B is an 8 billion parameter language model that combines masked diffusion with interpretable concept decomposition. Unlike standard autoregressive LLMs, Steerling generates text by iteratively unmasking tokens in order of confidence, and decomposes its internal representations into human-interpretable concepts that can be inspected and steered.
20
 
21
  ## Quick Start
@@ -98,4 +102,4 @@ these upstream licenses for downstream use of the model weights.
98
  Please check back for updates on the weight licensing terms.
99
 
100
  For questions about commercial use of the model weights,
101
- contact us at info@guidelabs.ai
 
1
  ---
 
 
 
 
 
 
 
 
2
  language:
3
+ - en
4
+ library_name: steerling
5
+ license: apache-2.0
6
  pipeline_tag: text-generation
7
+ tags:
8
+ - causal-diffusion
9
+ - interpretability
10
+ - concept-steering
11
+ - masked-diffusion
12
+ - block-causal
13
  ---
14
 
15
  # Steerling-8B:
16
 
17
  **An interpretable causal diffusion language model with concept steering.**
18
 
19
+ **Paper**: [Scaling Inherently Interpretable Language Models](https://huggingface.co/papers/2608.07594)
20
+ **Code**: [github.com/guidelabs/steerling](https://github.com/guidelabs/steerling)
21
+ **Project page**: [guidelabs.ai/papers/scaling-inherently-interpretable-language-models](https://www.guidelabs.ai/papers/scaling-inherently-interpretable-language-models/)
22
+
23
  Steerling-8B is an 8 billion parameter language model that combines masked diffusion with interpretable concept decomposition. Unlike standard autoregressive LLMs, Steerling generates text by iteratively unmasking tokens in order of confidence, and decomposes its internal representations into human-interpretable concepts that can be inspected and steered.
24
 
25
  ## Quick Start
 
102
  Please check back for updates on the weight licensing terms.
103
 
104
  For questions about commercial use of the model weights,
105
+ contact us at info@guidelabs.ai