wbit5 commited on
Commit
34abae0
Β·
verified Β·
1 Parent(s): d34f468

publish from LatentPortPort 473f8ef

Browse files
Files changed (1) hide show
  1. README.md +17 -1
README.md CHANGED
@@ -83,6 +83,21 @@ Gemma 3 270M reads; Gemma 3 1B answers, across sliding-window and global attenti
83
  βš–οΈ **Gated, and not Apache-2.0:** it is a Gemma Model Derivative, distributed under the
84
  [Gemma Terms of Use](https://ai.google.dev/gemma/terms).
85
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
86
  ### πŸ”€ A hybrid model β€” [`qwen3.5-0.8b-to-2b-bridge`](https://huggingface.co/agentic-underground/qwen3.5-0.8b-to-2b-bridge)
87
 
88
  Qwen3.5-0.8B reads; Qwen3.5-2B answers. Qwen3.5 mixes full-attention layers with Gated DeltaNet
@@ -124,7 +139,8 @@ and measured.
124
  - πŸ™ **Attributed** β€” PrattleNot is a derivative work of [LatentPort](https://latentport.com/) by
125
  Simon Villani ([source](https://github.com/LatentPort/latentport), Apache-2.0). Our bridges are
126
  Apache-2.0, except those fitted from Gemma, which are Gemma Model Derivatives, gated and distributed
127
- under the [Gemma Terms of Use](https://ai.google.dev/gemma/terms).
 
128
 
129
  ## Elsewhere
130
 
 
83
  βš–οΈ **Gated, and not Apache-2.0:** it is a Gemma Model Derivative, distributed under the
84
  [Gemma Terms of Use](https://ai.google.dev/gemma/terms).
85
 
86
+ ### πŸ¦™ A fourth family β€” [`llama-3.2-1b-to-3b-bridge`](https://huggingface.co/agentic-underground/llama-3.2-1b-to-3b-bridge)
87
+
88
+ **Built with Llama.** Llama 3.2 1B reads; Llama 3.2 3B answers, across different head widths (64
89
+ and 128).
90
+
91
+ | | |
92
+ |:--|:--|
93
+ | πŸ” **Fidelity (NCR)** | **0.965** with the paper's learned correction (0.818 from the maps alone), our highest yet |
94
+ | πŸ§ͺ **Control (wrong donor)** | βˆ’0.040 |
95
+ | 🎯 **Held-out questions** | **6 of 7**, the same as the 3B alone (the 1B: 7). No answer gain yet |
96
+ | ⚑ **Time to first token** | **1.70Γ—** sooner (95% CI 1.66–1.74) on one RTX 3090 at ~256 tokens; ceiling 2.54Γ— |
97
+
98
+ βš–οΈ **Gated, and not Apache-2.0:** it is a derivative work of Llama 3.2, distributed under the
99
+ [Llama 3.2 Community License](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct/blob/main/LICENSE.txt).
100
+
101
  ### πŸ”€ A hybrid model β€” [`qwen3.5-0.8b-to-2b-bridge`](https://huggingface.co/agentic-underground/qwen3.5-0.8b-to-2b-bridge)
102
 
103
  Qwen3.5-0.8B reads; Qwen3.5-2B answers. Qwen3.5 mixes full-attention layers with Gated DeltaNet
 
139
  - πŸ™ **Attributed** β€” PrattleNot is a derivative work of [LatentPort](https://latentport.com/) by
140
  Simon Villani ([source](https://github.com/LatentPort/latentport), Apache-2.0). Our bridges are
141
  Apache-2.0, except those fitted from Gemma, which are Gemma Model Derivatives, gated and distributed
142
+ under the [Gemma Terms of Use](https://ai.google.dev/gemma/terms), and those fitted from Llama, which
143
+ are Built with Llama, gated and distributed under the Llama 3.2 Community License.
144
 
145
  ## Elsewhere
146