Hugging Face's logo

AxiaoDBL
/
DeepSeek-R1-0528-Qwen3-8B-CodeLx-Reasoning

Transformers
Safetensors
Model card Files
xet
 Community
AxiaoDBL commited on
Commit
6e68005
·
verified ·
1 Parent(s): 8691c75

Upload model

Browse files
Files changed (2) hide show
  1. adapter_config.json +8 -6
  2. adapter_model.safetensors +1 -1
adapter_config.json CHANGED
@@ -20,20 +20,22 @@
20
  "megatron_core": "megatron.core",
21
  "modules_to_save": null,
22
  "peft_type": "LORA",
 
23
  "r": 64,
24
  "rank_pattern": {},
25
  "revision": null,
26
  "target_modules": [
27
- "up_proj",
28
- "v_proj",
29
- "q_proj",
30
- "o_proj",
31
- "gate_proj",
32
  "down_proj",
33
- "k_proj"
 
 
 
 
34
  ],
35
  "task_type": "CAUSAL_LM",
36
  "trainable_token_indices": null,
37
  "use_dora": false,
 
38
  "use_rslora": false
39
  }
 
20
  "megatron_core": "megatron.core",
21
  "modules_to_save": null,
22
  "peft_type": "LORA",
23
+ "qalora_group_size": 16,
24
  "r": 64,
25
  "rank_pattern": {},
26
  "revision": null,
27
  "target_modules": [
28
+ "k_proj",
 
 
 
 
29
  "down_proj",
30
+ "gate_proj",
31
+ "q_proj",
32
+ "v_proj",
33
+ "up_proj",
34
+ "o_proj"
35
  ],
36
  "task_type": "CAUSAL_LM",
37
  "trainable_token_indices": null,
38
  "use_dora": false,
39
+ "use_qalora": false,
40
  "use_rslora": false
41
  }
adapter_model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:4f67847a6eba56ecc8b996e46d9678385ddb7cf4aa43e137c4bf26ae406e0196
3
  size 698419728
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6ac9bae410dac962bae484226493d57cf776d2ac25c39cfd91372bf831767aa6
3
  size 698419728