Hugging Face's logo

jinliuxi
/
deepmini-it

Safetensors
Chinese
English
deepseek_v3
deepseek
Mixture of Experts
instruction-tuning
bilingual
reasoning
code
math
Model card Files Community
jinliuxi commited on
Commit
eb76c1b
·
verified ·
1 Parent(s): a022d31

Upload DeepseekV3ForCausalLM

Browse files
Files changed (2) hide show
  1. config.json +4 -4
  2. model.safetensors +1 -1
config.json CHANGED
@@ -1,5 +1,5 @@
1
  {
2
- "_name_or_path": "deepseekmini_highit/checkpoint-4961",
3
  "architectures": [
4
  "DeepseekV3ForCausalLM"
5
  ],
@@ -35,9 +35,9 @@
35
  "qk_rope_head_dim": 16,
36
  "rms_norm_eps": 1e-06,
37
  "rope_scaling": {
38
- "beta_fast": 32.0,
39
- "beta_slow": 1.0,
40
- "factor": 40.0,
41
  "mscale": 1.0,
42
  "mscale_all_dim": 1.0,
43
  "original_max_position_embeddings": 2048,
 
1
  {
2
+ "_name_or_path": "deepseekmini_refineit/checkpoint-3512",
3
  "architectures": [
4
  "DeepseekV3ForCausalLM"
5
  ],
 
35
  "qk_rope_head_dim": 16,
36
  "rms_norm_eps": 1e-06,
37
  "rope_scaling": {
38
+ "beta_fast": 32,
39
+ "beta_slow": 1,
40
+ "factor": 40,
41
  "mscale": 1.0,
42
  "mscale_all_dim": 1.0,
43
  "original_max_position_embeddings": 2048,
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:98fc5b5efcdf56b90674b33dec996f4dd9b93220c5115caf9d10c0c2b4fc8060
3
  size 1341455152
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bd5a5cf025668d2579f77925d315c4ffad066e5cdb926cf8b50c40883578766d
3
  size 1341455152