mrm8488 commited on
Commit
d34b565
·
verified ·
1 Parent(s): 88d49af

Upload model

Browse files
Files changed (2) hide show
  1. config.json +22 -1
  2. model.safetensors +2 -2
config.json CHANGED
@@ -23,5 +23,26 @@
23
  "transformers_version": "4.57.0",
24
  "type_vocab_size": 1,
25
  "use_cache": true,
26
- "vocab_size": 250002
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
27
  }
 
23
  "transformers_version": "4.57.0",
24
  "type_vocab_size": 1,
25
  "use_cache": true,
26
+ "vocab_size": 32768,
27
+ "vocabtrimmer": {
28
+ "mining_config": {
29
+ "dataset": "lbourdois/fineweb-2-trimming",
30
+ "dataset_column": "text",
31
+ "dataset_name": "spa",
32
+ "dataset_split": "train",
33
+ "language": "spa",
34
+ "min_frequency": 2,
35
+ "target_vocab_size": 32766
36
+ },
37
+ "stats": {
38
+ "compression_rate_embedding": 13.107095143238853,
39
+ "compression_rate_full": 60.24853369361119,
40
+ "parameter_size_embedding/raw": 256002048,
41
+ "parameter_size_embedding/trimmed": 33554432,
42
+ "parameter_size_full/raw": 560142482,
43
+ "parameter_size_full/trimmed": 337477632,
44
+ "vocab_size/raw": 250002,
45
+ "vocab_size/trimmed": 32768
46
+ }
47
+ }
48
  }
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:d0d642a915473631cdf27ac25ca63d7fdd75a3bc3ef6cac2202cbcd63043cf20
3
- size 2239607176
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0e2ea63e9f7da68262029ade5bb3fd7de1a757f1d26b10171d308aae376e5f4b
3
+ size 1349816152