TinyLamda commited on
Commit
4330c63
·
verified ·
1 Parent(s): e942f19

README: mask_locality layout entry, q3_8b license row

Browse files
Files changed (1) hide show
  1. README.md +2 -0
README.md CHANGED
@@ -14,6 +14,7 @@ license_name: per-model
14
  - `<model>/manifest.json`: file list, sha256 of every file and tensor, restore paths
15
  - `<model>/ppl/`: tokenized WikiText-2 / C4 evaluation windows (int32 token ids) with their `manifest.json`
16
  - `<model>/moments/`: calibration second-moment statistics the key-side bases were fitted from
 
17
  - **Format (.lqz)**: a 4-bit group of 128 values holds at most 16 distinct fp16 values, stored as a 16-entry fp16 table plus
18
  128 4-bit indices; all other tensors are stored raw. Decoding is exact (bit-identical tensors).
19
  - **Not** an inference-ready release: research artifacts only, no quality guarantee.
@@ -25,6 +26,7 @@ Each folder carries the license of the model it is derived from.
25
  | Folder | Derived from | License |
26
  |---|---|---|
27
  | `q3_4b/` | `Qwen/Qwen3-4B` | Apache-2.0 |
 
28
  | `l31i/` | `meta-llama/Llama-3.1-8B-Instruct` | Llama 3.1 Community License (`l31i/LICENSE`, `l31i/USE_POLICY.md`) |
29
  | `l31/` | `meta-llama/Llama-3.1-8B` | Llama 3.1 Community License (`l31/LICENSE`, `l31/USE_POLICY.md`) |
30
  | `l2/` | `meta-llama/Llama-2-7b-hf` | Llama 2 Community License (`l2/LICENSE.txt`, `l2/USE_POLICY.md`) |
 
14
  - `<model>/manifest.json`: file list, sha256 of every file and tensor, restore paths
15
  - `<model>/ppl/`: tokenized WikiText-2 / C4 evaluation windows (int32 token ids) with their `manifest.json`
16
  - `<model>/moments/`: calibration second-moment statistics the key-side bases were fitted from
17
+ - `<model>/mask_locality/cache/`: inputs and intermediate products of a pruning-mask locality study (channel permutations from calibration, per-channel keep-frequency profiles); sha256 of every file is recorded in the study's `ARTIFACTS.json`
18
  - **Format (.lqz)**: a 4-bit group of 128 values holds at most 16 distinct fp16 values, stored as a 16-entry fp16 table plus
19
  128 4-bit indices; all other tensors are stored raw. Decoding is exact (bit-identical tensors).
20
  - **Not** an inference-ready release: research artifacts only, no quality guarantee.
 
26
  | Folder | Derived from | License |
27
  |---|---|---|
28
  | `q3_4b/` | `Qwen/Qwen3-4B` | Apache-2.0 |
29
+ | `q3_8b/` | `Qwen/Qwen3-8B` | Apache-2.0 |
30
  | `l31i/` | `meta-llama/Llama-3.1-8B-Instruct` | Llama 3.1 Community License (`l31i/LICENSE`, `l31i/USE_POLICY.md`) |
31
  | `l31/` | `meta-llama/Llama-3.1-8B` | Llama 3.1 Community License (`l31/LICENSE`, `l31/USE_POLICY.md`) |
32
  | `l2/` | `meta-llama/Llama-2-7b-hf` | Llama 2 Community License (`l2/LICENSE.txt`, `l2/USE_POLICY.md`) |