Hugging Face's logo

DatPySci
/
RLVR-SGDM-Gap

Safetensors
Model card Files
xet
 Community
DatPySci commited on
Commit
10c9d5a
·
verified ·
1 Parent(s): caf63f2

upload lora

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_128/actor/lora_adapter/adapter_config.json +49 -0
  2. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_128/actor/lora_adapter/adapter_model.safetensors +3 -0
  3. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_192/actor/lora_adapter/adapter_config.json +49 -0
  4. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_192/actor/lora_adapter/adapter_model.safetensors +3 -0
  5. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_256/actor/lora_adapter/adapter_config.json +49 -0
  6. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_256/actor/lora_adapter/adapter_model.safetensors +3 -0
  7. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_320/actor/lora_adapter/adapter_config.json +49 -0
  8. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_320/actor/lora_adapter/adapter_model.safetensors +3 -0
  9. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_384/actor/lora_adapter/adapter_config.json +49 -0
  10. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_384/actor/lora_adapter/adapter_model.safetensors +3 -0
  11. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_448/actor/lora_adapter/adapter_config.json +49 -0
  12. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_448/actor/lora_adapter/adapter_model.safetensors +3 -0
  13. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_512/actor/lora_adapter/adapter_config.json +49 -0
  14. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_512/actor/lora_adapter/adapter_model.safetensors +3 -0
  15. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_64/actor/lora_adapter/adapter_config.json +49 -0
  16. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_64/actor/lora_adapter/adapter_model.safetensors +3 -0
  17. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_128/actor/lora_adapter/adapter_config.json +49 -0
  18. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_128/actor/lora_adapter/adapter_model.safetensors +3 -0
  19. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_192/actor/lora_adapter/adapter_config.json +49 -0
  20. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_192/actor/lora_adapter/adapter_model.safetensors +3 -0
  21. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_256/actor/lora_adapter/adapter_config.json +49 -0
  22. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_256/actor/lora_adapter/adapter_model.safetensors +3 -0
  23. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_320/actor/lora_adapter/adapter_config.json +49 -0
  24. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_320/actor/lora_adapter/adapter_model.safetensors +3 -0
  25. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_384/actor/lora_adapter/adapter_config.json +49 -0
  26. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_384/actor/lora_adapter/adapter_model.safetensors +3 -0
  27. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_448/actor/lora_adapter/adapter_config.json +49 -0
  28. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_448/actor/lora_adapter/adapter_model.safetensors +3 -0
  29. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_512/actor/lora_adapter/adapter_config.json +49 -0
  30. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_512/actor/lora_adapter/adapter_model.safetensors +3 -0
  31. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_64/actor/lora_adapter/adapter_config.json +49 -0
  32. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_64/actor/lora_adapter/adapter_model.safetensors +3 -0
  33. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_128/actor/lora_adapter/adapter_config.json +49 -0
  34. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_128/actor/lora_adapter/adapter_model.safetensors +3 -0
  35. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_192/actor/lora_adapter/adapter_config.json +49 -0
  36. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_192/actor/lora_adapter/adapter_model.safetensors +3 -0
  37. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_256/actor/lora_adapter/adapter_config.json +49 -0
  38. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_256/actor/lora_adapter/adapter_model.safetensors +3 -0
  39. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_320/actor/lora_adapter/adapter_config.json +49 -0
  40. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_320/actor/lora_adapter/adapter_model.safetensors +3 -0
  41. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_384/actor/lora_adapter/adapter_config.json +49 -0
  42. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_384/actor/lora_adapter/adapter_model.safetensors +3 -0
  43. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_448/actor/lora_adapter/adapter_config.json +49 -0
  44. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_448/actor/lora_adapter/adapter_model.safetensors +3 -0
  45. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_512/actor/lora_adapter/adapter_config.json +49 -0
  46. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_512/actor/lora_adapter/adapter_model.safetensors +3 -0
  47. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_64/actor/lora_adapter/adapter_config.json +49 -0
  48. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_64/actor/lora_adapter/adapter_model.safetensors +3 -0
  49. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank8/global_step_128/actor/lora_adapter/adapter_config.json +49 -0
  50. qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank8/global_step_128/actor/lora_adapter/adapter_model.safetensors +3 -0
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_128/actor/lora_adapter/adapter_config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "task_type": "CAUSAL_LM",
3
+ "peft_type": "LORA",
4
+ "auto_mapping": null,
5
+ "peft_version": "0.18.1",
6
+ "base_model_name_or_path": "/dev/shm/verl-cache/b3c28c1b99a08b84eb28d5733b49c01c/aa8e72537993ba99e69dfaafa59ed015b17504d1",
7
+ "revision": null,
8
+ "inference_mode": false,
9
+ "r": 16,
10
+ "target_modules": [
11
+ "down_proj",
12
+ "up_proj",
13
+ "q_proj",
14
+ "o_proj",
15
+ "gate_proj",
16
+ "k_proj",
17
+ "v_proj"
18
+ ],
19
+ "exclude_modules": null,
20
+ "lora_alpha": 32,
21
+ "lora_dropout": 0.0,
22
+ "fan_in_fan_out": false,
23
+ "bias": "none",
24
+ "use_rslora": false,
25
+ "modules_to_save": null,
26
+ "init_lora_weights": true,
27
+ "layers_to_transform": null,
28
+ "layers_pattern": null,
29
+ "rank_pattern": {},
30
+ "alpha_pattern": {},
31
+ "megatron_config": null,
32
+ "megatron_core": "megatron.core",
33
+ "trainable_token_indices": null,
34
+ "loftq_config": {},
35
+ "eva_config": null,
36
+ "corda_config": null,
37
+ "use_dora": false,
38
+ "alora_invocation_tokens": null,
39
+ "use_qalora": false,
40
+ "qalora_group_size": 16,
41
+ "layer_replication": null,
42
+ "runtime_config": {
43
+ "ephemeral_gpu_offload": false
44
+ },
45
+ "lora_bias": false,
46
+ "target_parameters": null,
47
+ "arrow_config": null,
48
+ "ensure_weight_tying": false
49
+ }
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_128/actor/lora_adapter/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4f822a3c589425866b8e43570095c11e50db20fcba48e4a5a0e7a235b7c40527
3
+ size 119801496
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_192/actor/lora_adapter/adapter_config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "task_type": "CAUSAL_LM",
3
+ "peft_type": "LORA",
4
+ "auto_mapping": null,
5
+ "peft_version": "0.18.1",
6
+ "base_model_name_or_path": "/dev/shm/verl-cache/b3c28c1b99a08b84eb28d5733b49c01c/aa8e72537993ba99e69dfaafa59ed015b17504d1",
7
+ "revision": null,
8
+ "inference_mode": false,
9
+ "r": 16,
10
+ "target_modules": [
11
+ "down_proj",
12
+ "up_proj",
13
+ "q_proj",
14
+ "o_proj",
15
+ "gate_proj",
16
+ "k_proj",
17
+ "v_proj"
18
+ ],
19
+ "exclude_modules": null,
20
+ "lora_alpha": 32,
21
+ "lora_dropout": 0.0,
22
+ "fan_in_fan_out": false,
23
+ "bias": "none",
24
+ "use_rslora": false,
25
+ "modules_to_save": null,
26
+ "init_lora_weights": true,
27
+ "layers_to_transform": null,
28
+ "layers_pattern": null,
29
+ "rank_pattern": {},
30
+ "alpha_pattern": {},
31
+ "megatron_config": null,
32
+ "megatron_core": "megatron.core",
33
+ "trainable_token_indices": null,
34
+ "loftq_config": {},
35
+ "eva_config": null,
36
+ "corda_config": null,
37
+ "use_dora": false,
38
+ "alora_invocation_tokens": null,
39
+ "use_qalora": false,
40
+ "qalora_group_size": 16,
41
+ "layer_replication": null,
42
+ "runtime_config": {
43
+ "ephemeral_gpu_offload": false
44
+ },
45
+ "lora_bias": false,
46
+ "target_parameters": null,
47
+ "arrow_config": null,
48
+ "ensure_weight_tying": false
49
+ }
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_192/actor/lora_adapter/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f552989eece9f876e5767a916e82da6f6d8037a2cfe1276ea0c5b2d59a6b4977
3
+ size 119801496
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_256/actor/lora_adapter/adapter_config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "task_type": "CAUSAL_LM",
3
+ "peft_type": "LORA",
4
+ "auto_mapping": null,
5
+ "peft_version": "0.18.1",
6
+ "base_model_name_or_path": "/dev/shm/verl-cache/b3c28c1b99a08b84eb28d5733b49c01c/aa8e72537993ba99e69dfaafa59ed015b17504d1",
7
+ "revision": null,
8
+ "inference_mode": false,
9
+ "r": 16,
10
+ "target_modules": [
11
+ "down_proj",
12
+ "up_proj",
13
+ "q_proj",
14
+ "o_proj",
15
+ "gate_proj",
16
+ "k_proj",
17
+ "v_proj"
18
+ ],
19
+ "exclude_modules": null,
20
+ "lora_alpha": 32,
21
+ "lora_dropout": 0.0,
22
+ "fan_in_fan_out": false,
23
+ "bias": "none",
24
+ "use_rslora": false,
25
+ "modules_to_save": null,
26
+ "init_lora_weights": true,
27
+ "layers_to_transform": null,
28
+ "layers_pattern": null,
29
+ "rank_pattern": {},
30
+ "alpha_pattern": {},
31
+ "megatron_config": null,
32
+ "megatron_core": "megatron.core",
33
+ "trainable_token_indices": null,
34
+ "loftq_config": {},
35
+ "eva_config": null,
36
+ "corda_config": null,
37
+ "use_dora": false,
38
+ "alora_invocation_tokens": null,
39
+ "use_qalora": false,
40
+ "qalora_group_size": 16,
41
+ "layer_replication": null,
42
+ "runtime_config": {
43
+ "ephemeral_gpu_offload": false
44
+ },
45
+ "lora_bias": false,
46
+ "target_parameters": null,
47
+ "arrow_config": null,
48
+ "ensure_weight_tying": false
49
+ }
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_256/actor/lora_adapter/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:209855544cc22542ddf222d1ca8f421bf107ceaca61e678191a99b65d5fff3cf
3
+ size 119801496
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_320/actor/lora_adapter/adapter_config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "task_type": "CAUSAL_LM",
3
+ "peft_type": "LORA",
4
+ "auto_mapping": null,
5
+ "peft_version": "0.18.1",
6
+ "base_model_name_or_path": "/dev/shm/verl-cache/b3c28c1b99a08b84eb28d5733b49c01c/aa8e72537993ba99e69dfaafa59ed015b17504d1",
7
+ "revision": null,
8
+ "inference_mode": false,
9
+ "r": 16,
10
+ "target_modules": [
11
+ "down_proj",
12
+ "up_proj",
13
+ "q_proj",
14
+ "o_proj",
15
+ "gate_proj",
16
+ "k_proj",
17
+ "v_proj"
18
+ ],
19
+ "exclude_modules": null,
20
+ "lora_alpha": 32,
21
+ "lora_dropout": 0.0,
22
+ "fan_in_fan_out": false,
23
+ "bias": "none",
24
+ "use_rslora": false,
25
+ "modules_to_save": null,
26
+ "init_lora_weights": true,
27
+ "layers_to_transform": null,
28
+ "layers_pattern": null,
29
+ "rank_pattern": {},
30
+ "alpha_pattern": {},
31
+ "megatron_config": null,
32
+ "megatron_core": "megatron.core",
33
+ "trainable_token_indices": null,
34
+ "loftq_config": {},
35
+ "eva_config": null,
36
+ "corda_config": null,
37
+ "use_dora": false,
38
+ "alora_invocation_tokens": null,
39
+ "use_qalora": false,
40
+ "qalora_group_size": 16,
41
+ "layer_replication": null,
42
+ "runtime_config": {
43
+ "ephemeral_gpu_offload": false
44
+ },
45
+ "lora_bias": false,
46
+ "target_parameters": null,
47
+ "arrow_config": null,
48
+ "ensure_weight_tying": false
49
+ }
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_320/actor/lora_adapter/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2e5953d8c176e4bbf04085616f43e7ee8872553e393d7e6281779c2b7cf6fbca
3
+ size 119801496
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_384/actor/lora_adapter/adapter_config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "task_type": "CAUSAL_LM",
3
+ "peft_type": "LORA",
4
+ "auto_mapping": null,
5
+ "peft_version": "0.18.1",
6
+ "base_model_name_or_path": "/dev/shm/verl-cache/b3c28c1b99a08b84eb28d5733b49c01c/aa8e72537993ba99e69dfaafa59ed015b17504d1",
7
+ "revision": null,
8
+ "inference_mode": false,
9
+ "r": 16,
10
+ "target_modules": [
11
+ "down_proj",
12
+ "up_proj",
13
+ "q_proj",
14
+ "o_proj",
15
+ "gate_proj",
16
+ "k_proj",
17
+ "v_proj"
18
+ ],
19
+ "exclude_modules": null,
20
+ "lora_alpha": 32,
21
+ "lora_dropout": 0.0,
22
+ "fan_in_fan_out": false,
23
+ "bias": "none",
24
+ "use_rslora": false,
25
+ "modules_to_save": null,
26
+ "init_lora_weights": true,
27
+ "layers_to_transform": null,
28
+ "layers_pattern": null,
29
+ "rank_pattern": {},
30
+ "alpha_pattern": {},
31
+ "megatron_config": null,
32
+ "megatron_core": "megatron.core",
33
+ "trainable_token_indices": null,
34
+ "loftq_config": {},
35
+ "eva_config": null,
36
+ "corda_config": null,
37
+ "use_dora": false,
38
+ "alora_invocation_tokens": null,
39
+ "use_qalora": false,
40
+ "qalora_group_size": 16,
41
+ "layer_replication": null,
42
+ "runtime_config": {
43
+ "ephemeral_gpu_offload": false
44
+ },
45
+ "lora_bias": false,
46
+ "target_parameters": null,
47
+ "arrow_config": null,
48
+ "ensure_weight_tying": false
49
+ }
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_384/actor/lora_adapter/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eb67622d2db0eda89f89c15d5107d6f79b83697c5714e2866ca481986e029ecf
3
+ size 119801496
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_448/actor/lora_adapter/adapter_config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "task_type": "CAUSAL_LM",
3
+ "peft_type": "LORA",
4
+ "auto_mapping": null,
5
+ "peft_version": "0.18.1",
6
+ "base_model_name_or_path": "/dev/shm/verl-cache/b3c28c1b99a08b84eb28d5733b49c01c/aa8e72537993ba99e69dfaafa59ed015b17504d1",
7
+ "revision": null,
8
+ "inference_mode": false,
9
+ "r": 16,
10
+ "target_modules": [
11
+ "down_proj",
12
+ "up_proj",
13
+ "q_proj",
14
+ "o_proj",
15
+ "gate_proj",
16
+ "k_proj",
17
+ "v_proj"
18
+ ],
19
+ "exclude_modules": null,
20
+ "lora_alpha": 32,
21
+ "lora_dropout": 0.0,
22
+ "fan_in_fan_out": false,
23
+ "bias": "none",
24
+ "use_rslora": false,
25
+ "modules_to_save": null,
26
+ "init_lora_weights": true,
27
+ "layers_to_transform": null,
28
+ "layers_pattern": null,
29
+ "rank_pattern": {},
30
+ "alpha_pattern": {},
31
+ "megatron_config": null,
32
+ "megatron_core": "megatron.core",
33
+ "trainable_token_indices": null,
34
+ "loftq_config": {},
35
+ "eva_config": null,
36
+ "corda_config": null,
37
+ "use_dora": false,
38
+ "alora_invocation_tokens": null,
39
+ "use_qalora": false,
40
+ "qalora_group_size": 16,
41
+ "layer_replication": null,
42
+ "runtime_config": {
43
+ "ephemeral_gpu_offload": false
44
+ },
45
+ "lora_bias": false,
46
+ "target_parameters": null,
47
+ "arrow_config": null,
48
+ "ensure_weight_tying": false
49
+ }
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_448/actor/lora_adapter/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4b58e383eabfdf47630c4e95c903b9929efe156acebc80b668b5bf296040cad9
3
+ size 119801496
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_512/actor/lora_adapter/adapter_config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "task_type": "CAUSAL_LM",
3
+ "peft_type": "LORA",
4
+ "auto_mapping": null,
5
+ "peft_version": "0.18.1",
6
+ "base_model_name_or_path": "/dev/shm/verl-cache/b3c28c1b99a08b84eb28d5733b49c01c/aa8e72537993ba99e69dfaafa59ed015b17504d1",
7
+ "revision": null,
8
+ "inference_mode": false,
9
+ "r": 16,
10
+ "target_modules": [
11
+ "down_proj",
12
+ "up_proj",
13
+ "q_proj",
14
+ "o_proj",
15
+ "gate_proj",
16
+ "k_proj",
17
+ "v_proj"
18
+ ],
19
+ "exclude_modules": null,
20
+ "lora_alpha": 32,
21
+ "lora_dropout": 0.0,
22
+ "fan_in_fan_out": false,
23
+ "bias": "none",
24
+ "use_rslora": false,
25
+ "modules_to_save": null,
26
+ "init_lora_weights": true,
27
+ "layers_to_transform": null,
28
+ "layers_pattern": null,
29
+ "rank_pattern": {},
30
+ "alpha_pattern": {},
31
+ "megatron_config": null,
32
+ "megatron_core": "megatron.core",
33
+ "trainable_token_indices": null,
34
+ "loftq_config": {},
35
+ "eva_config": null,
36
+ "corda_config": null,
37
+ "use_dora": false,
38
+ "alora_invocation_tokens": null,
39
+ "use_qalora": false,
40
+ "qalora_group_size": 16,
41
+ "layer_replication": null,
42
+ "runtime_config": {
43
+ "ephemeral_gpu_offload": false
44
+ },
45
+ "lora_bias": false,
46
+ "target_parameters": null,
47
+ "arrow_config": null,
48
+ "ensure_weight_tying": false
49
+ }
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_512/actor/lora_adapter/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:00bec09157f2ee7dbce109867a22419be66fd7a570c9ae7df6349338632173e7
3
+ size 119801496
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_64/actor/lora_adapter/adapter_config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "task_type": "CAUSAL_LM",
3
+ "peft_type": "LORA",
4
+ "auto_mapping": null,
5
+ "peft_version": "0.18.1",
6
+ "base_model_name_or_path": "/dev/shm/verl-cache/b3c28c1b99a08b84eb28d5733b49c01c/aa8e72537993ba99e69dfaafa59ed015b17504d1",
7
+ "revision": null,
8
+ "inference_mode": false,
9
+ "r": 16,
10
+ "target_modules": [
11
+ "down_proj",
12
+ "up_proj",
13
+ "q_proj",
14
+ "o_proj",
15
+ "gate_proj",
16
+ "k_proj",
17
+ "v_proj"
18
+ ],
19
+ "exclude_modules": null,
20
+ "lora_alpha": 32,
21
+ "lora_dropout": 0.0,
22
+ "fan_in_fan_out": false,
23
+ "bias": "none",
24
+ "use_rslora": false,
25
+ "modules_to_save": null,
26
+ "init_lora_weights": true,
27
+ "layers_to_transform": null,
28
+ "layers_pattern": null,
29
+ "rank_pattern": {},
30
+ "alpha_pattern": {},
31
+ "megatron_config": null,
32
+ "megatron_core": "megatron.core",
33
+ "trainable_token_indices": null,
34
+ "loftq_config": {},
35
+ "eva_config": null,
36
+ "corda_config": null,
37
+ "use_dora": false,
38
+ "alora_invocation_tokens": null,
39
+ "use_qalora": false,
40
+ "qalora_group_size": 16,
41
+ "layer_replication": null,
42
+ "runtime_config": {
43
+ "ephemeral_gpu_offload": false
44
+ },
45
+ "lora_bias": false,
46
+ "target_parameters": null,
47
+ "arrow_config": null,
48
+ "ensure_weight_tying": false
49
+ }
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank16/global_step_64/actor/lora_adapter/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ecbe1f57a24c8880112a4f30c52682930041296406739a8ef3875b673dce9320
3
+ size 119801496
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_128/actor/lora_adapter/adapter_config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "task_type": "CAUSAL_LM",
3
+ "peft_type": "LORA",
4
+ "auto_mapping": null,
5
+ "peft_version": "0.18.1",
6
+ "base_model_name_or_path": "/dev/shm/verl-cache/b3c28c1b99a08b84eb28d5733b49c01c/aa8e72537993ba99e69dfaafa59ed015b17504d1",
7
+ "revision": null,
8
+ "inference_mode": false,
9
+ "r": 32,
10
+ "target_modules": [
11
+ "gate_proj",
12
+ "down_proj",
13
+ "q_proj",
14
+ "k_proj",
15
+ "up_proj",
16
+ "o_proj",
17
+ "v_proj"
18
+ ],
19
+ "exclude_modules": null,
20
+ "lora_alpha": 64,
21
+ "lora_dropout": 0.0,
22
+ "fan_in_fan_out": false,
23
+ "bias": "none",
24
+ "use_rslora": false,
25
+ "modules_to_save": null,
26
+ "init_lora_weights": true,
27
+ "layers_to_transform": null,
28
+ "layers_pattern": null,
29
+ "rank_pattern": {},
30
+ "alpha_pattern": {},
31
+ "megatron_config": null,
32
+ "megatron_core": "megatron.core",
33
+ "trainable_token_indices": null,
34
+ "loftq_config": {},
35
+ "eva_config": null,
36
+ "corda_config": null,
37
+ "use_dora": false,
38
+ "alora_invocation_tokens": null,
39
+ "use_qalora": false,
40
+ "qalora_group_size": 16,
41
+ "layer_replication": null,
42
+ "runtime_config": {
43
+ "ephemeral_gpu_offload": false
44
+ },
45
+ "lora_bias": false,
46
+ "target_parameters": null,
47
+ "arrow_config": null,
48
+ "ensure_weight_tying": false
49
+ }
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_128/actor/lora_adapter/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8679bb391f69a1a137426889ecd2a592d673ce4461fab94ca19af74e9bf64c39
3
+ size 239536248
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_192/actor/lora_adapter/adapter_config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "task_type": "CAUSAL_LM",
3
+ "peft_type": "LORA",
4
+ "auto_mapping": null,
5
+ "peft_version": "0.18.1",
6
+ "base_model_name_or_path": "/dev/shm/verl-cache/b3c28c1b99a08b84eb28d5733b49c01c/aa8e72537993ba99e69dfaafa59ed015b17504d1",
7
+ "revision": null,
8
+ "inference_mode": false,
9
+ "r": 32,
10
+ "target_modules": [
11
+ "gate_proj",
12
+ "down_proj",
13
+ "q_proj",
14
+ "k_proj",
15
+ "up_proj",
16
+ "o_proj",
17
+ "v_proj"
18
+ ],
19
+ "exclude_modules": null,
20
+ "lora_alpha": 64,
21
+ "lora_dropout": 0.0,
22
+ "fan_in_fan_out": false,
23
+ "bias": "none",
24
+ "use_rslora": false,
25
+ "modules_to_save": null,
26
+ "init_lora_weights": true,
27
+ "layers_to_transform": null,
28
+ "layers_pattern": null,
29
+ "rank_pattern": {},
30
+ "alpha_pattern": {},
31
+ "megatron_config": null,
32
+ "megatron_core": "megatron.core",
33
+ "trainable_token_indices": null,
34
+ "loftq_config": {},
35
+ "eva_config": null,
36
+ "corda_config": null,
37
+ "use_dora": false,
38
+ "alora_invocation_tokens": null,
39
+ "use_qalora": false,
40
+ "qalora_group_size": 16,
41
+ "layer_replication": null,
42
+ "runtime_config": {
43
+ "ephemeral_gpu_offload": false
44
+ },
45
+ "lora_bias": false,
46
+ "target_parameters": null,
47
+ "arrow_config": null,
48
+ "ensure_weight_tying": false
49
+ }
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_192/actor/lora_adapter/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:665c58e6c5367d50ceef9e419ece12469ead9c6b5944d84195ee11d5e7ee99cc
3
+ size 239536248
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_256/actor/lora_adapter/adapter_config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "task_type": "CAUSAL_LM",
3
+ "peft_type": "LORA",
4
+ "auto_mapping": null,
5
+ "peft_version": "0.18.1",
6
+ "base_model_name_or_path": "/dev/shm/verl-cache/b3c28c1b99a08b84eb28d5733b49c01c/aa8e72537993ba99e69dfaafa59ed015b17504d1",
7
+ "revision": null,
8
+ "inference_mode": false,
9
+ "r": 32,
10
+ "target_modules": [
11
+ "gate_proj",
12
+ "down_proj",
13
+ "q_proj",
14
+ "k_proj",
15
+ "up_proj",
16
+ "o_proj",
17
+ "v_proj"
18
+ ],
19
+ "exclude_modules": null,
20
+ "lora_alpha": 64,
21
+ "lora_dropout": 0.0,
22
+ "fan_in_fan_out": false,
23
+ "bias": "none",
24
+ "use_rslora": false,
25
+ "modules_to_save": null,
26
+ "init_lora_weights": true,
27
+ "layers_to_transform": null,
28
+ "layers_pattern": null,
29
+ "rank_pattern": {},
30
+ "alpha_pattern": {},
31
+ "megatron_config": null,
32
+ "megatron_core": "megatron.core",
33
+ "trainable_token_indices": null,
34
+ "loftq_config": {},
35
+ "eva_config": null,
36
+ "corda_config": null,
37
+ "use_dora": false,
38
+ "alora_invocation_tokens": null,
39
+ "use_qalora": false,
40
+ "qalora_group_size": 16,
41
+ "layer_replication": null,
42
+ "runtime_config": {
43
+ "ephemeral_gpu_offload": false
44
+ },
45
+ "lora_bias": false,
46
+ "target_parameters": null,
47
+ "arrow_config": null,
48
+ "ensure_weight_tying": false
49
+ }
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_256/actor/lora_adapter/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b567bbd1225cee5dcfbc1e8de6f541011d67f6d3f729d2dbd98b5661c5f08da6
3
+ size 239536248
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_320/actor/lora_adapter/adapter_config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "task_type": "CAUSAL_LM",
3
+ "peft_type": "LORA",
4
+ "auto_mapping": null,
5
+ "peft_version": "0.18.1",
6
+ "base_model_name_or_path": "/dev/shm/verl-cache/b3c28c1b99a08b84eb28d5733b49c01c/aa8e72537993ba99e69dfaafa59ed015b17504d1",
7
+ "revision": null,
8
+ "inference_mode": false,
9
+ "r": 32,
10
+ "target_modules": [
11
+ "gate_proj",
12
+ "down_proj",
13
+ "q_proj",
14
+ "k_proj",
15
+ "up_proj",
16
+ "o_proj",
17
+ "v_proj"
18
+ ],
19
+ "exclude_modules": null,
20
+ "lora_alpha": 64,
21
+ "lora_dropout": 0.0,
22
+ "fan_in_fan_out": false,
23
+ "bias": "none",
24
+ "use_rslora": false,
25
+ "modules_to_save": null,
26
+ "init_lora_weights": true,
27
+ "layers_to_transform": null,
28
+ "layers_pattern": null,
29
+ "rank_pattern": {},
30
+ "alpha_pattern": {},
31
+ "megatron_config": null,
32
+ "megatron_core": "megatron.core",
33
+ "trainable_token_indices": null,
34
+ "loftq_config": {},
35
+ "eva_config": null,
36
+ "corda_config": null,
37
+ "use_dora": false,
38
+ "alora_invocation_tokens": null,
39
+ "use_qalora": false,
40
+ "qalora_group_size": 16,
41
+ "layer_replication": null,
42
+ "runtime_config": {
43
+ "ephemeral_gpu_offload": false
44
+ },
45
+ "lora_bias": false,
46
+ "target_parameters": null,
47
+ "arrow_config": null,
48
+ "ensure_weight_tying": false
49
+ }
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_320/actor/lora_adapter/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3e941eec492fefda234a9da89b908c7fda18ba9ae3b5108e29cb723abf7e191a
3
+ size 239536248
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_384/actor/lora_adapter/adapter_config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "task_type": "CAUSAL_LM",
3
+ "peft_type": "LORA",
4
+ "auto_mapping": null,
5
+ "peft_version": "0.18.1",
6
+ "base_model_name_or_path": "/dev/shm/verl-cache/b3c28c1b99a08b84eb28d5733b49c01c/aa8e72537993ba99e69dfaafa59ed015b17504d1",
7
+ "revision": null,
8
+ "inference_mode": false,
9
+ "r": 32,
10
+ "target_modules": [
11
+ "gate_proj",
12
+ "down_proj",
13
+ "q_proj",
14
+ "k_proj",
15
+ "up_proj",
16
+ "o_proj",
17
+ "v_proj"
18
+ ],
19
+ "exclude_modules": null,
20
+ "lora_alpha": 64,
21
+ "lora_dropout": 0.0,
22
+ "fan_in_fan_out": false,
23
+ "bias": "none",
24
+ "use_rslora": false,
25
+ "modules_to_save": null,
26
+ "init_lora_weights": true,
27
+ "layers_to_transform": null,
28
+ "layers_pattern": null,
29
+ "rank_pattern": {},
30
+ "alpha_pattern": {},
31
+ "megatron_config": null,
32
+ "megatron_core": "megatron.core",
33
+ "trainable_token_indices": null,
34
+ "loftq_config": {},
35
+ "eva_config": null,
36
+ "corda_config": null,
37
+ "use_dora": false,
38
+ "alora_invocation_tokens": null,
39
+ "use_qalora": false,
40
+ "qalora_group_size": 16,
41
+ "layer_replication": null,
42
+ "runtime_config": {
43
+ "ephemeral_gpu_offload": false
44
+ },
45
+ "lora_bias": false,
46
+ "target_parameters": null,
47
+ "arrow_config": null,
48
+ "ensure_weight_tying": false
49
+ }
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_384/actor/lora_adapter/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b7118e7a2f7010c18b62ce0f6bd8ed7e9657373e5a3728cb825feedd79e1b4b2
3
+ size 239536248
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_448/actor/lora_adapter/adapter_config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "task_type": "CAUSAL_LM",
3
+ "peft_type": "LORA",
4
+ "auto_mapping": null,
5
+ "peft_version": "0.18.1",
6
+ "base_model_name_or_path": "/dev/shm/verl-cache/b3c28c1b99a08b84eb28d5733b49c01c/aa8e72537993ba99e69dfaafa59ed015b17504d1",
7
+ "revision": null,
8
+ "inference_mode": false,
9
+ "r": 32,
10
+ "target_modules": [
11
+ "gate_proj",
12
+ "down_proj",
13
+ "q_proj",
14
+ "k_proj",
15
+ "up_proj",
16
+ "o_proj",
17
+ "v_proj"
18
+ ],
19
+ "exclude_modules": null,
20
+ "lora_alpha": 64,
21
+ "lora_dropout": 0.0,
22
+ "fan_in_fan_out": false,
23
+ "bias": "none",
24
+ "use_rslora": false,
25
+ "modules_to_save": null,
26
+ "init_lora_weights": true,
27
+ "layers_to_transform": null,
28
+ "layers_pattern": null,
29
+ "rank_pattern": {},
30
+ "alpha_pattern": {},
31
+ "megatron_config": null,
32
+ "megatron_core": "megatron.core",
33
+ "trainable_token_indices": null,
34
+ "loftq_config": {},
35
+ "eva_config": null,
36
+ "corda_config": null,
37
+ "use_dora": false,
38
+ "alora_invocation_tokens": null,
39
+ "use_qalora": false,
40
+ "qalora_group_size": 16,
41
+ "layer_replication": null,
42
+ "runtime_config": {
43
+ "ephemeral_gpu_offload": false
44
+ },
45
+ "lora_bias": false,
46
+ "target_parameters": null,
47
+ "arrow_config": null,
48
+ "ensure_weight_tying": false
49
+ }
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_448/actor/lora_adapter/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:620bd7650389babaf6e5076e1c2675f6c95d9138d9c317a2b91215bc978ea65b
3
+ size 239536248
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_512/actor/lora_adapter/adapter_config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "task_type": "CAUSAL_LM",
3
+ "peft_type": "LORA",
4
+ "auto_mapping": null,
5
+ "peft_version": "0.18.1",
6
+ "base_model_name_or_path": "/dev/shm/verl-cache/b3c28c1b99a08b84eb28d5733b49c01c/aa8e72537993ba99e69dfaafa59ed015b17504d1",
7
+ "revision": null,
8
+ "inference_mode": false,
9
+ "r": 32,
10
+ "target_modules": [
11
+ "gate_proj",
12
+ "down_proj",
13
+ "q_proj",
14
+ "k_proj",
15
+ "up_proj",
16
+ "o_proj",
17
+ "v_proj"
18
+ ],
19
+ "exclude_modules": null,
20
+ "lora_alpha": 64,
21
+ "lora_dropout": 0.0,
22
+ "fan_in_fan_out": false,
23
+ "bias": "none",
24
+ "use_rslora": false,
25
+ "modules_to_save": null,
26
+ "init_lora_weights": true,
27
+ "layers_to_transform": null,
28
+ "layers_pattern": null,
29
+ "rank_pattern": {},
30
+ "alpha_pattern": {},
31
+ "megatron_config": null,
32
+ "megatron_core": "megatron.core",
33
+ "trainable_token_indices": null,
34
+ "loftq_config": {},
35
+ "eva_config": null,
36
+ "corda_config": null,
37
+ "use_dora": false,
38
+ "alora_invocation_tokens": null,
39
+ "use_qalora": false,
40
+ "qalora_group_size": 16,
41
+ "layer_replication": null,
42
+ "runtime_config": {
43
+ "ephemeral_gpu_offload": false
44
+ },
45
+ "lora_bias": false,
46
+ "target_parameters": null,
47
+ "arrow_config": null,
48
+ "ensure_weight_tying": false
49
+ }
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_512/actor/lora_adapter/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0d1378877f1380822865cd69e3aa520b4653e57f2744e60a57be5ae86a627a6c
3
+ size 239536248
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_64/actor/lora_adapter/adapter_config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "task_type": "CAUSAL_LM",
3
+ "peft_type": "LORA",
4
+ "auto_mapping": null,
5
+ "peft_version": "0.18.1",
6
+ "base_model_name_or_path": "/dev/shm/verl-cache/b3c28c1b99a08b84eb28d5733b49c01c/aa8e72537993ba99e69dfaafa59ed015b17504d1",
7
+ "revision": null,
8
+ "inference_mode": false,
9
+ "r": 32,
10
+ "target_modules": [
11
+ "gate_proj",
12
+ "down_proj",
13
+ "q_proj",
14
+ "k_proj",
15
+ "up_proj",
16
+ "o_proj",
17
+ "v_proj"
18
+ ],
19
+ "exclude_modules": null,
20
+ "lora_alpha": 64,
21
+ "lora_dropout": 0.0,
22
+ "fan_in_fan_out": false,
23
+ "bias": "none",
24
+ "use_rslora": false,
25
+ "modules_to_save": null,
26
+ "init_lora_weights": true,
27
+ "layers_to_transform": null,
28
+ "layers_pattern": null,
29
+ "rank_pattern": {},
30
+ "alpha_pattern": {},
31
+ "megatron_config": null,
32
+ "megatron_core": "megatron.core",
33
+ "trainable_token_indices": null,
34
+ "loftq_config": {},
35
+ "eva_config": null,
36
+ "corda_config": null,
37
+ "use_dora": false,
38
+ "alora_invocation_tokens": null,
39
+ "use_qalora": false,
40
+ "qalora_group_size": 16,
41
+ "layer_replication": null,
42
+ "runtime_config": {
43
+ "ephemeral_gpu_offload": false
44
+ },
45
+ "lora_bias": false,
46
+ "target_parameters": null,
47
+ "arrow_config": null,
48
+ "ensure_weight_tying": false
49
+ }
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank32/global_step_64/actor/lora_adapter/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9507ee9b3ca871711aea06b38f44e40fbac7392133a6a2222ede283f4c1376f0
3
+ size 239536248
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_128/actor/lora_adapter/adapter_config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "task_type": "CAUSAL_LM",
3
+ "peft_type": "LORA",
4
+ "auto_mapping": null,
5
+ "peft_version": "0.18.1",
6
+ "base_model_name_or_path": "/dev/shm/verl-cache/b3c28c1b99a08b84eb28d5733b49c01c/aa8e72537993ba99e69dfaafa59ed015b17504d1",
7
+ "revision": null,
8
+ "inference_mode": false,
9
+ "r": 64,
10
+ "target_modules": [
11
+ "gate_proj",
12
+ "k_proj",
13
+ "v_proj",
14
+ "down_proj",
15
+ "q_proj",
16
+ "up_proj",
17
+ "o_proj"
18
+ ],
19
+ "exclude_modules": null,
20
+ "lora_alpha": 128,
21
+ "lora_dropout": 0.0,
22
+ "fan_in_fan_out": false,
23
+ "bias": "none",
24
+ "use_rslora": false,
25
+ "modules_to_save": null,
26
+ "init_lora_weights": true,
27
+ "layers_to_transform": null,
28
+ "layers_pattern": null,
29
+ "rank_pattern": {},
30
+ "alpha_pattern": {},
31
+ "megatron_config": null,
32
+ "megatron_core": "megatron.core",
33
+ "trainable_token_indices": null,
34
+ "loftq_config": {},
35
+ "eva_config": null,
36
+ "corda_config": null,
37
+ "use_dora": false,
38
+ "alora_invocation_tokens": null,
39
+ "use_qalora": false,
40
+ "qalora_group_size": 16,
41
+ "layer_replication": null,
42
+ "runtime_config": {
43
+ "ephemeral_gpu_offload": false
44
+ },
45
+ "lora_bias": false,
46
+ "target_parameters": null,
47
+ "arrow_config": null,
48
+ "ensure_weight_tying": false
49
+ }
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_128/actor/lora_adapter/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2d560e44898365c51a026af8774bed301c94201088a500971f98b8c9701e91bd
3
+ size 479005032
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_192/actor/lora_adapter/adapter_config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "task_type": "CAUSAL_LM",
3
+ "peft_type": "LORA",
4
+ "auto_mapping": null,
5
+ "peft_version": "0.18.1",
6
+ "base_model_name_or_path": "/dev/shm/verl-cache/b3c28c1b99a08b84eb28d5733b49c01c/aa8e72537993ba99e69dfaafa59ed015b17504d1",
7
+ "revision": null,
8
+ "inference_mode": false,
9
+ "r": 64,
10
+ "target_modules": [
11
+ "gate_proj",
12
+ "k_proj",
13
+ "v_proj",
14
+ "down_proj",
15
+ "q_proj",
16
+ "up_proj",
17
+ "o_proj"
18
+ ],
19
+ "exclude_modules": null,
20
+ "lora_alpha": 128,
21
+ "lora_dropout": 0.0,
22
+ "fan_in_fan_out": false,
23
+ "bias": "none",
24
+ "use_rslora": false,
25
+ "modules_to_save": null,
26
+ "init_lora_weights": true,
27
+ "layers_to_transform": null,
28
+ "layers_pattern": null,
29
+ "rank_pattern": {},
30
+ "alpha_pattern": {},
31
+ "megatron_config": null,
32
+ "megatron_core": "megatron.core",
33
+ "trainable_token_indices": null,
34
+ "loftq_config": {},
35
+ "eva_config": null,
36
+ "corda_config": null,
37
+ "use_dora": false,
38
+ "alora_invocation_tokens": null,
39
+ "use_qalora": false,
40
+ "qalora_group_size": 16,
41
+ "layer_replication": null,
42
+ "runtime_config": {
43
+ "ephemeral_gpu_offload": false
44
+ },
45
+ "lora_bias": false,
46
+ "target_parameters": null,
47
+ "arrow_config": null,
48
+ "ensure_weight_tying": false
49
+ }
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_192/actor/lora_adapter/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5950026a5e2bdcbd20f764182ddf53aafbb9a9a38d50daf6b92fe5f41210828f
3
+ size 479005032
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_256/actor/lora_adapter/adapter_config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "task_type": "CAUSAL_LM",
3
+ "peft_type": "LORA",
4
+ "auto_mapping": null,
5
+ "peft_version": "0.18.1",
6
+ "base_model_name_or_path": "/dev/shm/verl-cache/b3c28c1b99a08b84eb28d5733b49c01c/aa8e72537993ba99e69dfaafa59ed015b17504d1",
7
+ "revision": null,
8
+ "inference_mode": false,
9
+ "r": 64,
10
+ "target_modules": [
11
+ "gate_proj",
12
+ "k_proj",
13
+ "v_proj",
14
+ "down_proj",
15
+ "q_proj",
16
+ "up_proj",
17
+ "o_proj"
18
+ ],
19
+ "exclude_modules": null,
20
+ "lora_alpha": 128,
21
+ "lora_dropout": 0.0,
22
+ "fan_in_fan_out": false,
23
+ "bias": "none",
24
+ "use_rslora": false,
25
+ "modules_to_save": null,
26
+ "init_lora_weights": true,
27
+ "layers_to_transform": null,
28
+ "layers_pattern": null,
29
+ "rank_pattern": {},
30
+ "alpha_pattern": {},
31
+ "megatron_config": null,
32
+ "megatron_core": "megatron.core",
33
+ "trainable_token_indices": null,
34
+ "loftq_config": {},
35
+ "eva_config": null,
36
+ "corda_config": null,
37
+ "use_dora": false,
38
+ "alora_invocation_tokens": null,
39
+ "use_qalora": false,
40
+ "qalora_group_size": 16,
41
+ "layer_replication": null,
42
+ "runtime_config": {
43
+ "ephemeral_gpu_offload": false
44
+ },
45
+ "lora_bias": false,
46
+ "target_parameters": null,
47
+ "arrow_config": null,
48
+ "ensure_weight_tying": false
49
+ }
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_256/actor/lora_adapter/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:29d3dd11e2f379d49325d4cfb88c9ce50c42ee3c2857c070e7d1b363fd59666d
3
+ size 479005032
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_320/actor/lora_adapter/adapter_config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "task_type": "CAUSAL_LM",
3
+ "peft_type": "LORA",
4
+ "auto_mapping": null,
5
+ "peft_version": "0.18.1",
6
+ "base_model_name_or_path": "/dev/shm/verl-cache/b3c28c1b99a08b84eb28d5733b49c01c/aa8e72537993ba99e69dfaafa59ed015b17504d1",
7
+ "revision": null,
8
+ "inference_mode": false,
9
+ "r": 64,
10
+ "target_modules": [
11
+ "gate_proj",
12
+ "k_proj",
13
+ "v_proj",
14
+ "down_proj",
15
+ "q_proj",
16
+ "up_proj",
17
+ "o_proj"
18
+ ],
19
+ "exclude_modules": null,
20
+ "lora_alpha": 128,
21
+ "lora_dropout": 0.0,
22
+ "fan_in_fan_out": false,
23
+ "bias": "none",
24
+ "use_rslora": false,
25
+ "modules_to_save": null,
26
+ "init_lora_weights": true,
27
+ "layers_to_transform": null,
28
+ "layers_pattern": null,
29
+ "rank_pattern": {},
30
+ "alpha_pattern": {},
31
+ "megatron_config": null,
32
+ "megatron_core": "megatron.core",
33
+ "trainable_token_indices": null,
34
+ "loftq_config": {},
35
+ "eva_config": null,
36
+ "corda_config": null,
37
+ "use_dora": false,
38
+ "alora_invocation_tokens": null,
39
+ "use_qalora": false,
40
+ "qalora_group_size": 16,
41
+ "layer_replication": null,
42
+ "runtime_config": {
43
+ "ephemeral_gpu_offload": false
44
+ },
45
+ "lora_bias": false,
46
+ "target_parameters": null,
47
+ "arrow_config": null,
48
+ "ensure_weight_tying": false
49
+ }
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_320/actor/lora_adapter/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5c77eeb1471d65ac214f4ec52ed3343ca093145712ffe4fbf946694dc7fac838
3
+ size 479005032
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_384/actor/lora_adapter/adapter_config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "task_type": "CAUSAL_LM",
3
+ "peft_type": "LORA",
4
+ "auto_mapping": null,
5
+ "peft_version": "0.18.1",
6
+ "base_model_name_or_path": "/dev/shm/verl-cache/b3c28c1b99a08b84eb28d5733b49c01c/aa8e72537993ba99e69dfaafa59ed015b17504d1",
7
+ "revision": null,
8
+ "inference_mode": false,
9
+ "r": 64,
10
+ "target_modules": [
11
+ "gate_proj",
12
+ "k_proj",
13
+ "v_proj",
14
+ "down_proj",
15
+ "q_proj",
16
+ "up_proj",
17
+ "o_proj"
18
+ ],
19
+ "exclude_modules": null,
20
+ "lora_alpha": 128,
21
+ "lora_dropout": 0.0,
22
+ "fan_in_fan_out": false,
23
+ "bias": "none",
24
+ "use_rslora": false,
25
+ "modules_to_save": null,
26
+ "init_lora_weights": true,
27
+ "layers_to_transform": null,
28
+ "layers_pattern": null,
29
+ "rank_pattern": {},
30
+ "alpha_pattern": {},
31
+ "megatron_config": null,
32
+ "megatron_core": "megatron.core",
33
+ "trainable_token_indices": null,
34
+ "loftq_config": {},
35
+ "eva_config": null,
36
+ "corda_config": null,
37
+ "use_dora": false,
38
+ "alora_invocation_tokens": null,
39
+ "use_qalora": false,
40
+ "qalora_group_size": 16,
41
+ "layer_replication": null,
42
+ "runtime_config": {
43
+ "ephemeral_gpu_offload": false
44
+ },
45
+ "lora_bias": false,
46
+ "target_parameters": null,
47
+ "arrow_config": null,
48
+ "ensure_weight_tying": false
49
+ }
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_384/actor/lora_adapter/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7c7f5a8736495fca81a8121790f780718aade8683e965c1911c8dfe0a2ada6b4
3
+ size 479005032
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_448/actor/lora_adapter/adapter_config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "task_type": "CAUSAL_LM",
3
+ "peft_type": "LORA",
4
+ "auto_mapping": null,
5
+ "peft_version": "0.18.1",
6
+ "base_model_name_or_path": "/dev/shm/verl-cache/b3c28c1b99a08b84eb28d5733b49c01c/aa8e72537993ba99e69dfaafa59ed015b17504d1",
7
+ "revision": null,
8
+ "inference_mode": false,
9
+ "r": 64,
10
+ "target_modules": [
11
+ "gate_proj",
12
+ "k_proj",
13
+ "v_proj",
14
+ "down_proj",
15
+ "q_proj",
16
+ "up_proj",
17
+ "o_proj"
18
+ ],
19
+ "exclude_modules": null,
20
+ "lora_alpha": 128,
21
+ "lora_dropout": 0.0,
22
+ "fan_in_fan_out": false,
23
+ "bias": "none",
24
+ "use_rslora": false,
25
+ "modules_to_save": null,
26
+ "init_lora_weights": true,
27
+ "layers_to_transform": null,
28
+ "layers_pattern": null,
29
+ "rank_pattern": {},
30
+ "alpha_pattern": {},
31
+ "megatron_config": null,
32
+ "megatron_core": "megatron.core",
33
+ "trainable_token_indices": null,
34
+ "loftq_config": {},
35
+ "eva_config": null,
36
+ "corda_config": null,
37
+ "use_dora": false,
38
+ "alora_invocation_tokens": null,
39
+ "use_qalora": false,
40
+ "qalora_group_size": 16,
41
+ "layer_replication": null,
42
+ "runtime_config": {
43
+ "ephemeral_gpu_offload": false
44
+ },
45
+ "lora_bias": false,
46
+ "target_parameters": null,
47
+ "arrow_config": null,
48
+ "ensure_weight_tying": false
49
+ }
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_448/actor/lora_adapter/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1f19a4b8876b799ba6d699d3ea4ad887c57995f5c08c89590b6f966eeaae07e4
3
+ size 479005032
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_512/actor/lora_adapter/adapter_config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "task_type": "CAUSAL_LM",
3
+ "peft_type": "LORA",
4
+ "auto_mapping": null,
5
+ "peft_version": "0.18.1",
6
+ "base_model_name_or_path": "/dev/shm/verl-cache/b3c28c1b99a08b84eb28d5733b49c01c/aa8e72537993ba99e69dfaafa59ed015b17504d1",
7
+ "revision": null,
8
+ "inference_mode": false,
9
+ "r": 64,
10
+ "target_modules": [
11
+ "gate_proj",
12
+ "k_proj",
13
+ "v_proj",
14
+ "down_proj",
15
+ "q_proj",
16
+ "up_proj",
17
+ "o_proj"
18
+ ],
19
+ "exclude_modules": null,
20
+ "lora_alpha": 128,
21
+ "lora_dropout": 0.0,
22
+ "fan_in_fan_out": false,
23
+ "bias": "none",
24
+ "use_rslora": false,
25
+ "modules_to_save": null,
26
+ "init_lora_weights": true,
27
+ "layers_to_transform": null,
28
+ "layers_pattern": null,
29
+ "rank_pattern": {},
30
+ "alpha_pattern": {},
31
+ "megatron_config": null,
32
+ "megatron_core": "megatron.core",
33
+ "trainable_token_indices": null,
34
+ "loftq_config": {},
35
+ "eva_config": null,
36
+ "corda_config": null,
37
+ "use_dora": false,
38
+ "alora_invocation_tokens": null,
39
+ "use_qalora": false,
40
+ "qalora_group_size": 16,
41
+ "layer_replication": null,
42
+ "runtime_config": {
43
+ "ephemeral_gpu_offload": false
44
+ },
45
+ "lora_bias": false,
46
+ "target_parameters": null,
47
+ "arrow_config": null,
48
+ "ensure_weight_tying": false
49
+ }
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_512/actor/lora_adapter/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:59d0459898f25508e73b421d187806d3ec076d4b321d7c31e54c7074bc3d6754
3
+ size 479005032
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_64/actor/lora_adapter/adapter_config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "task_type": "CAUSAL_LM",
3
+ "peft_type": "LORA",
4
+ "auto_mapping": null,
5
+ "peft_version": "0.18.1",
6
+ "base_model_name_or_path": "/dev/shm/verl-cache/b3c28c1b99a08b84eb28d5733b49c01c/aa8e72537993ba99e69dfaafa59ed015b17504d1",
7
+ "revision": null,
8
+ "inference_mode": false,
9
+ "r": 64,
10
+ "target_modules": [
11
+ "gate_proj",
12
+ "k_proj",
13
+ "v_proj",
14
+ "down_proj",
15
+ "q_proj",
16
+ "up_proj",
17
+ "o_proj"
18
+ ],
19
+ "exclude_modules": null,
20
+ "lora_alpha": 128,
21
+ "lora_dropout": 0.0,
22
+ "fan_in_fan_out": false,
23
+ "bias": "none",
24
+ "use_rslora": false,
25
+ "modules_to_save": null,
26
+ "init_lora_weights": true,
27
+ "layers_to_transform": null,
28
+ "layers_pattern": null,
29
+ "rank_pattern": {},
30
+ "alpha_pattern": {},
31
+ "megatron_config": null,
32
+ "megatron_core": "megatron.core",
33
+ "trainable_token_indices": null,
34
+ "loftq_config": {},
35
+ "eva_config": null,
36
+ "corda_config": null,
37
+ "use_dora": false,
38
+ "alora_invocation_tokens": null,
39
+ "use_qalora": false,
40
+ "qalora_group_size": 16,
41
+ "layer_replication": null,
42
+ "runtime_config": {
43
+ "ephemeral_gpu_offload": false
44
+ },
45
+ "lora_bias": false,
46
+ "target_parameters": null,
47
+ "arrow_config": null,
48
+ "ensure_weight_tying": false
49
+ }
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank64/global_step_64/actor/lora_adapter/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:066386917aeb9d85b6963d30b3100a8d62dcb4b49fd9ab2d4cc3bd12ff5bb2b6
3
+ size 479005032
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank8/global_step_128/actor/lora_adapter/adapter_config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "task_type": "CAUSAL_LM",
3
+ "peft_type": "LORA",
4
+ "auto_mapping": null,
5
+ "peft_version": "0.18.1",
6
+ "base_model_name_or_path": "/dev/shm/verl-cache/b3c28c1b99a08b84eb28d5733b49c01c/aa8e72537993ba99e69dfaafa59ed015b17504d1",
7
+ "revision": null,
8
+ "inference_mode": false,
9
+ "r": 8,
10
+ "target_modules": [
11
+ "down_proj",
12
+ "gate_proj",
13
+ "k_proj",
14
+ "up_proj",
15
+ "q_proj",
16
+ "o_proj",
17
+ "v_proj"
18
+ ],
19
+ "exclude_modules": null,
20
+ "lora_alpha": 16,
21
+ "lora_dropout": 0.0,
22
+ "fan_in_fan_out": false,
23
+ "bias": "none",
24
+ "use_rslora": false,
25
+ "modules_to_save": null,
26
+ "init_lora_weights": true,
27
+ "layers_to_transform": null,
28
+ "layers_pattern": null,
29
+ "rank_pattern": {},
30
+ "alpha_pattern": {},
31
+ "megatron_config": null,
32
+ "megatron_core": "megatron.core",
33
+ "trainable_token_indices": null,
34
+ "loftq_config": {},
35
+ "eva_config": null,
36
+ "corda_config": null,
37
+ "use_dora": false,
38
+ "alora_invocation_tokens": null,
39
+ "use_qalora": false,
40
+ "qalora_group_size": 16,
41
+ "layer_replication": null,
42
+ "runtime_config": {
43
+ "ephemeral_gpu_offload": false
44
+ },
45
+ "lora_bias": false,
46
+ "target_parameters": null,
47
+ "arrow_config": null,
48
+ "ensure_weight_tying": false
49
+ }
qwen-3b-lora/Qwen/Qwen2.5-3B-Instruct-math-GRPO-LoRA-rank8/global_step_128/actor/lora_adapter/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:14e2e6ad8893cb2eea88c870d5e2c36f89a7adc6ce069cb458a155f8723cafa4
3
+ size 59933600