Hugging Face's logo

codecodebear
/
llm_source_balancing_sft_lora

Safetensors
Model card Files
xet
 Community
codecodebear commited on
Commit
b474e68
·
verified ·
1 Parent(s): b368f56

Add files using upload-large-folder tool

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. .gitattributes +6 -0
  2. csqa__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_0.pth +3 -0
  3. csqa__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_1.pth +3 -0
  4. csqa__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_2.pth +3 -0
  5. csqa__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_3.pth +3 -0
  6. csqa__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_4.pth +3 -0
  7. csqa__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_5.pth +3 -0
  8. csqa__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_6.pth +3 -0
  9. csqa__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_7.pth +3 -0
  10. csqa__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/scheduler.pt +3 -0
  11. csqa__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/tokenizer.json +3 -0
  12. csqa__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/training_args.bin +3 -0
  13. csqa__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/tokenizer.json +3 -0
  14. csqa__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/training_args.bin +3 -0
  15. csqa__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_0.pth +3 -0
  16. csqa__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_1.pth +3 -0
  17. csqa__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_2.pth +3 -0
  18. csqa__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_3.pth +3 -0
  19. csqa__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_4.pth +3 -0
  20. csqa__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_5.pth +3 -0
  21. csqa__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_6.pth +3 -0
  22. csqa__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_7.pth +3 -0
  23. csqa__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/scheduler.pt +3 -0
  24. csqa__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/tokenizer.json +3 -0
  25. csqa__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/training_args.bin +3 -0
  26. csqa__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/tokenizer.json +3 -0
  27. csqa__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/training_args.bin +3 -0
  28. csqa__qwen3_8b__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_0.pth +3 -0
  29. csqa__qwen3_8b__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_1.pth +3 -0
  30. csqa__qwen3_8b__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_2.pth +3 -0
  31. csqa__qwen3_8b__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_3.pth +3 -0
  32. csqa__qwen3_8b__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_4.pth +3 -0
  33. csqa__qwen3_8b__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_5.pth +3 -0
  34. csqa__qwen3_8b__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_6.pth +3 -0
  35. csqa__qwen3_8b__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_7.pth +3 -0
  36. csqa__qwen3_8b__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/scheduler.pt +3 -0
  37. csqa__qwen3_8b__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/tokenizer.json +3 -0
  38. csqa__qwen3_8b__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/training_args.bin +3 -0
  39. csqa__qwen3_8b__all_variants_r8_bs4_lr1e5_e3/lora/sft/training_args.bin +3 -0
  40. csqa__qwen3_8b__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/tokenizer.json +3 -0
  41. csqa__qwen3_8b__bare_100_r8_bs4_lr1e5_e3/lora/sft/training_args.bin +3 -0
  42. gsm8k__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_4.pth +3 -0
  43. gsm8k__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/training_args.bin +3 -0
  44. gsm8k__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_0.pth +3 -0
  45. gsm8k__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_1.pth +3 -0
  46. gsm8k__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_2.pth +3 -0
  47. gsm8k__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_3.pth +3 -0
  48. gsm8k__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_4.pth +3 -0
  49. gsm8k__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_5.pth +3 -0
  50. gsm8k__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_7.pth +3 -0
.gitattributes CHANGED
@@ -103,3 +103,9 @@ gsm8k__qwen3_8b__all_variants_r8_bs4_lr1e5_e3/lora/sft/tokenizer.json filter=lfs
103
  gsm8k__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/tokenizer.json filter=lfs diff=lfs merge=lfs -text
104
  csqa__qwen3_8b__all_variants_r8_bs4_lr1e5_e3/lora/sft/tokenizer.json filter=lfs diff=lfs merge=lfs -text
105
  csqa__qwen3_8b__bare_100_r8_bs4_lr1e5_e3/lora/sft/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 
 
 
 
 
 
 
103
  gsm8k__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/tokenizer.json filter=lfs diff=lfs merge=lfs -text
104
  csqa__qwen3_8b__all_variants_r8_bs4_lr1e5_e3/lora/sft/tokenizer.json filter=lfs diff=lfs merge=lfs -text
105
  csqa__qwen3_8b__bare_100_r8_bs4_lr1e5_e3/lora/sft/tokenizer.json filter=lfs diff=lfs merge=lfs -text
106
+ csqa__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/tokenizer.json filter=lfs diff=lfs merge=lfs -text
107
+ csqa__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/tokenizer.json filter=lfs diff=lfs merge=lfs -text
108
+ csqa__qwen3_8b__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/tokenizer.json filter=lfs diff=lfs merge=lfs -text
109
+ csqa__qwen3_8b__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/tokenizer.json filter=lfs diff=lfs merge=lfs -text
110
+ csqa__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/tokenizer.json filter=lfs diff=lfs merge=lfs -text
111
+ csqa__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/tokenizer.json filter=lfs diff=lfs merge=lfs -text
csqa__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f0d51a89f706019e82a1bece244691bbef31c0dc9544212f828d427b8052b077
3
+ size 16389
csqa__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9db2a9d890a3bdfb6477301adcccedc994df442a5cf1f4f617b4737086e632ef
3
+ size 16389
csqa__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:74f99367832c29a5aff6042c0dc14d87a3ffc16ce9cc6466678227421d213c43
3
+ size 16389
csqa__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2984cd1dfb7f3fdddf5cc99bb1dab2759a14eef425d83154d76b310151f5d01c
3
+ size 16389
csqa__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4cb1eac572046e66de7e9c52bba4d61d758728542e4fdc6d8b194a6c29e10f96
3
+ size 16389
csqa__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:41da76e42d56dcd9ee4821dcdf66fd5f2e82e821081ec3e07cb7e1e952647992
3
+ size 16389
csqa__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b1ef332606e64d74e7d5d444ae550c6edf2fe161722c5d652fd47845e2aa1894
3
+ size 16389
csqa__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f4b125a8a8f1fcea8543fd480628a947fbdff2ec9452d6a7be7e77a426d76b0c
3
+ size 16389
csqa__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:70583c9a51d736f1353b66ad07c780d71300daef1dfeacad27c4a2efb4fd9d68
3
+ size 1465
csqa__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8c1dcab308e7cf5970ea38815e0a62887d705c5b436f869ca27a5dcdd40c36a6
3
+ size 17210148
csqa__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fdc1b3bc49e1e979a001baf2ca82a5f042bdc27bd295bc8237d9f5c39026b8fd
3
+ size 6225
csqa__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8c1dcab308e7cf5970ea38815e0a62887d705c5b436f869ca27a5dcdd40c36a6
3
+ size 17210148
csqa__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fdc1b3bc49e1e979a001baf2ca82a5f042bdc27bd295bc8237d9f5c39026b8fd
3
+ size 6225
csqa__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f0d51a89f706019e82a1bece244691bbef31c0dc9544212f828d427b8052b077
3
+ size 16389
csqa__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9db2a9d890a3bdfb6477301adcccedc994df442a5cf1f4f617b4737086e632ef
3
+ size 16389
csqa__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:74f99367832c29a5aff6042c0dc14d87a3ffc16ce9cc6466678227421d213c43
3
+ size 16389
csqa__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2984cd1dfb7f3fdddf5cc99bb1dab2759a14eef425d83154d76b310151f5d01c
3
+ size 16389
csqa__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4cb1eac572046e66de7e9c52bba4d61d758728542e4fdc6d8b194a6c29e10f96
3
+ size 16389
csqa__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:41da76e42d56dcd9ee4821dcdf66fd5f2e82e821081ec3e07cb7e1e952647992
3
+ size 16389
csqa__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b1ef332606e64d74e7d5d444ae550c6edf2fe161722c5d652fd47845e2aa1894
3
+ size 16389
csqa__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f4b125a8a8f1fcea8543fd480628a947fbdff2ec9452d6a7be7e77a426d76b0c
3
+ size 16389
csqa__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:70583c9a51d736f1353b66ad07c780d71300daef1dfeacad27c4a2efb4fd9d68
3
+ size 1465
csqa__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8c1dcab308e7cf5970ea38815e0a62887d705c5b436f869ca27a5dcdd40c36a6
3
+ size 17210148
csqa__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c7765c5b78f375dcc02c499a3caae283f561d92654cc152d633b21dc10b0b9e1
3
+ size 6225
csqa__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8c1dcab308e7cf5970ea38815e0a62887d705c5b436f869ca27a5dcdd40c36a6
3
+ size 17210148
csqa__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c7765c5b78f375dcc02c499a3caae283f561d92654cc152d633b21dc10b0b9e1
3
+ size 6225
csqa__qwen3_8b__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f0d51a89f706019e82a1bece244691bbef31c0dc9544212f828d427b8052b077
3
+ size 16389
csqa__qwen3_8b__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9db2a9d890a3bdfb6477301adcccedc994df442a5cf1f4f617b4737086e632ef
3
+ size 16389
csqa__qwen3_8b__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:74f99367832c29a5aff6042c0dc14d87a3ffc16ce9cc6466678227421d213c43
3
+ size 16389
csqa__qwen3_8b__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2984cd1dfb7f3fdddf5cc99bb1dab2759a14eef425d83154d76b310151f5d01c
3
+ size 16389
csqa__qwen3_8b__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4cb1eac572046e66de7e9c52bba4d61d758728542e4fdc6d8b194a6c29e10f96
3
+ size 16389
csqa__qwen3_8b__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:41da76e42d56dcd9ee4821dcdf66fd5f2e82e821081ec3e07cb7e1e952647992
3
+ size 16389
csqa__qwen3_8b__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b1ef332606e64d74e7d5d444ae550c6edf2fe161722c5d652fd47845e2aa1894
3
+ size 16389
csqa__qwen3_8b__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f4b125a8a8f1fcea8543fd480628a947fbdff2ec9452d6a7be7e77a426d76b0c
3
+ size 16389
csqa__qwen3_8b__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:70583c9a51d736f1353b66ad07c780d71300daef1dfeacad27c4a2efb4fd9d68
3
+ size 1465
csqa__qwen3_8b__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aeb13307a71acd8fe81861d94ad54ab689df773318809eed3cbe794b4492dae4
3
+ size 11422654
csqa__qwen3_8b__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:18faf0b870b38c912c75e1ce9afdbe19af1c3fadd92209c7a90c663c6da06db4
3
+ size 6161
csqa__qwen3_8b__all_variants_r8_bs4_lr1e5_e3/lora/sft/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:18faf0b870b38c912c75e1ce9afdbe19af1c3fadd92209c7a90c663c6da06db4
3
+ size 6161
csqa__qwen3_8b__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aeb13307a71acd8fe81861d94ad54ab689df773318809eed3cbe794b4492dae4
3
+ size 11422654
csqa__qwen3_8b__bare_100_r8_bs4_lr1e5_e3/lora/sft/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8659b993a9025fced8eda027b3eb2c8012015107d38086de64143bfb5704ef31
3
+ size 6161
gsm8k__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4cb1eac572046e66de7e9c52bba4d61d758728542e4fdc6d8b194a6c29e10f96
3
+ size 16389
gsm8k__llama3_8b_instruct__all_variants_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1d818ffd17d6503309f5c3b497030d0e3a41785b1e06d81ccae31f09c44a5ba6
3
+ size 6225
gsm8k__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f0d51a89f706019e82a1bece244691bbef31c0dc9544212f828d427b8052b077
3
+ size 16389
gsm8k__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9db2a9d890a3bdfb6477301adcccedc994df442a5cf1f4f617b4737086e632ef
3
+ size 16389
gsm8k__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:74f99367832c29a5aff6042c0dc14d87a3ffc16ce9cc6466678227421d213c43
3
+ size 16389
gsm8k__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2984cd1dfb7f3fdddf5cc99bb1dab2759a14eef425d83154d76b310151f5d01c
3
+ size 16389
gsm8k__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4cb1eac572046e66de7e9c52bba4d61d758728542e4fdc6d8b194a6c29e10f96
3
+ size 16389
gsm8k__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:41da76e42d56dcd9ee4821dcdf66fd5f2e82e821081ec3e07cb7e1e952647992
3
+ size 16389
gsm8k__llama3_8b_instruct__bare_100_r8_bs4_lr1e5_e3/lora/sft/checkpoint-939/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f4b125a8a8f1fcea8543fd480628a947fbdff2ec9452d6a7be7e77a426d76b0c
3
+ size 16389