Commit
•
86c7ad8
1
Parent(s):
572e560
Training in progress epoch 3
Browse files- README.md +13 -0
- Whisper PEFT Fine-Tuning/events.out.tfevents.1696436338.MLbox.300106.0 +2 -2
- adapter_config.json +93 -93
- adapter_model.bin +2 -2
- best_checkpoint/README.md +91 -0
- best_checkpoint/adapter_config.json +178 -178
- best_checkpoint/adapter_model.bin +2 -2
- best_checkpoint/optimizer.bin +1 -1
- best_checkpoint/random_states_0.pkl +1 -1
- best_checkpoint/random_states_1.pkl +1 -1
- best_checkpoint/scheduler.bin +1 -1
README.md
CHANGED
|
@@ -54,6 +54,18 @@ The following `bitsandbytes` quantization config was used during training:
|
|
| 54 |
- bnb_4bit_use_double_quant: False
|
| 55 |
- bnb_4bit_compute_dtype: float32
|
| 56 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 57 |
The following `bitsandbytes` quantization config was used during training:
|
| 58 |
- quant_method: bitsandbytes
|
| 59 |
- load_in_8bit: True
|
|
@@ -84,6 +96,7 @@ The following `bitsandbytes` quantization config was used during training:
|
|
| 84 |
- PEFT 0.5.0
|
| 85 |
- PEFT 0.5.0
|
| 86 |
- PEFT 0.5.0
|
|
|
|
| 87 |
|
| 88 |
- PEFT 0.5.0.dev0
|
| 89 |
`bitsandbytes` quantization config was used during training:
|
|
|
|
| 54 |
- bnb_4bit_use_double_quant: False
|
| 55 |
- bnb_4bit_compute_dtype: float32
|
| 56 |
|
| 57 |
+
The following `bitsandbytes` quantization config was used during training:
|
| 58 |
+
- quant_method: bitsandbytes
|
| 59 |
+
- load_in_8bit: True
|
| 60 |
+
- load_in_4bit: False
|
| 61 |
+
- llm_int8_threshold: 6.0
|
| 62 |
+
- llm_int8_skip_modules: None
|
| 63 |
+
- llm_int8_enable_fp32_cpu_offload: False
|
| 64 |
+
- llm_int8_has_fp16_weight: False
|
| 65 |
+
- bnb_4bit_quant_type: fp4
|
| 66 |
+
- bnb_4bit_use_double_quant: False
|
| 67 |
+
- bnb_4bit_compute_dtype: float32
|
| 68 |
+
|
| 69 |
The following `bitsandbytes` quantization config was used during training:
|
| 70 |
- quant_method: bitsandbytes
|
| 71 |
- load_in_8bit: True
|
|
|
|
| 96 |
- PEFT 0.5.0
|
| 97 |
- PEFT 0.5.0
|
| 98 |
- PEFT 0.5.0
|
| 99 |
+
- PEFT 0.5.0
|
| 100 |
|
| 101 |
- PEFT 0.5.0.dev0
|
| 102 |
`bitsandbytes` quantization config was used during training:
|
Whisper PEFT Fine-Tuning/events.out.tfevents.1696436338.MLbox.300106.0
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b344924cb79c25b3e962e7d68421e413fa83e4d68b8e84f6481d5e17e53a6ffb
|
| 3 |
+
size 4069
|
adapter_config.json
CHANGED
|
@@ -210,12 +210,12 @@
|
|
| 210 |
true,
|
| 211 |
false,
|
| 212 |
false,
|
| 213 |
-
true,
|
| 214 |
false,
|
| 215 |
false,
|
| 216 |
false,
|
| 217 |
false,
|
| 218 |
-
|
|
|
|
| 219 |
],
|
| 220 |
"model.decoder.layers.1.fc1.lora_E": [
|
| 221 |
true,
|
|
@@ -365,7 +365,7 @@
|
|
| 365 |
true,
|
| 366 |
true,
|
| 367 |
true,
|
| 368 |
-
|
| 369 |
true,
|
| 370 |
true,
|
| 371 |
true,
|
|
@@ -549,7 +549,7 @@
|
|
| 549 |
false,
|
| 550 |
false,
|
| 551 |
false,
|
| 552 |
-
|
| 553 |
true,
|
| 554 |
true
|
| 555 |
],
|
|
@@ -574,7 +574,7 @@
|
|
| 574 |
false,
|
| 575 |
false,
|
| 576 |
false,
|
| 577 |
-
|
| 578 |
false,
|
| 579 |
false,
|
| 580 |
false,
|
|
@@ -656,7 +656,7 @@
|
|
| 656 |
true,
|
| 657 |
true,
|
| 658 |
true,
|
| 659 |
-
|
| 660 |
true,
|
| 661 |
true,
|
| 662 |
true,
|
|
@@ -1156,7 +1156,7 @@
|
|
| 1156 |
false
|
| 1157 |
],
|
| 1158 |
"model.decoder.layers.16.encoder_attn.out_proj.lora_E": [
|
| 1159 |
-
|
| 1160 |
true,
|
| 1161 |
false,
|
| 1162 |
false,
|
|
@@ -1296,7 +1296,7 @@
|
|
| 1296 |
false
|
| 1297 |
],
|
| 1298 |
"model.decoder.layers.17.encoder_attn.out_proj.lora_E": [
|
| 1299 |
-
|
| 1300 |
false,
|
| 1301 |
true,
|
| 1302 |
false,
|
|
@@ -1306,7 +1306,7 @@
|
|
| 1306 |
false,
|
| 1307 |
false,
|
| 1308 |
false,
|
| 1309 |
-
|
| 1310 |
false
|
| 1311 |
],
|
| 1312 |
"model.decoder.layers.17.encoder_attn.q_proj.lora_E": [
|
|
@@ -1461,7 +1461,7 @@
|
|
| 1461 |
false,
|
| 1462 |
false,
|
| 1463 |
false,
|
| 1464 |
-
|
| 1465 |
],
|
| 1466 |
"model.decoder.layers.18.encoder_attn.v_proj.lora_E": [
|
| 1467 |
true,
|
|
@@ -1521,7 +1521,7 @@
|
|
| 1521 |
],
|
| 1522 |
"model.decoder.layers.18.self_attn.out_proj.lora_E": [
|
| 1523 |
true,
|
| 1524 |
-
|
| 1525 |
true,
|
| 1526 |
false,
|
| 1527 |
true,
|
|
@@ -1529,7 +1529,7 @@
|
|
| 1529 |
true,
|
| 1530 |
false,
|
| 1531 |
true,
|
| 1532 |
-
|
| 1533 |
true,
|
| 1534 |
true
|
| 1535 |
],
|
|
@@ -1577,13 +1577,13 @@
|
|
| 1577 |
],
|
| 1578 |
"model.decoder.layers.19.encoder_attn.out_proj.lora_E": [
|
| 1579 |
false,
|
| 1580 |
-
true,
|
| 1581 |
false,
|
| 1582 |
false,
|
| 1583 |
false,
|
| 1584 |
false,
|
| 1585 |
false,
|
| 1586 |
-
|
|
|
|
| 1587 |
false,
|
| 1588 |
false,
|
| 1589 |
false,
|
|
@@ -1668,7 +1668,7 @@
|
|
| 1668 |
true,
|
| 1669 |
false,
|
| 1670 |
false,
|
| 1671 |
-
|
| 1672 |
true,
|
| 1673 |
false,
|
| 1674 |
false
|
|
@@ -1699,7 +1699,7 @@
|
|
| 1699 |
false,
|
| 1700 |
true,
|
| 1701 |
false,
|
| 1702 |
-
|
| 1703 |
],
|
| 1704 |
"model.decoder.layers.2.encoder_attn.k_proj.lora_E": [
|
| 1705 |
false,
|
|
@@ -1800,18 +1800,18 @@
|
|
| 1800 |
false
|
| 1801 |
],
|
| 1802 |
"model.decoder.layers.2.self_attn.out_proj.lora_E": [
|
| 1803 |
-
true,
|
| 1804 |
true,
|
| 1805 |
false,
|
| 1806 |
false,
|
| 1807 |
-
true,
|
| 1808 |
-
true,
|
| 1809 |
false,
|
| 1810 |
true,
|
|
|
|
|
|
|
|
|
|
| 1811 |
true,
|
| 1812 |
true,
|
| 1813 |
true,
|
| 1814 |
-
|
| 1815 |
],
|
| 1816 |
"model.decoder.layers.2.self_attn.q_proj.lora_E": [
|
| 1817 |
false,
|
|
@@ -1863,7 +1863,7 @@
|
|
| 1863 |
true,
|
| 1864 |
false,
|
| 1865 |
true,
|
| 1866 |
-
|
| 1867 |
true,
|
| 1868 |
true,
|
| 1869 |
true,
|
|
@@ -1922,7 +1922,7 @@
|
|
| 1922 |
true,
|
| 1923 |
true,
|
| 1924 |
true,
|
| 1925 |
-
|
| 1926 |
true
|
| 1927 |
],
|
| 1928 |
"model.decoder.layers.20.self_attn.k_proj.lora_E": [
|
|
@@ -1942,7 +1942,7 @@
|
|
| 1942 |
"model.decoder.layers.20.self_attn.out_proj.lora_E": [
|
| 1943 |
true,
|
| 1944 |
false,
|
| 1945 |
-
|
| 1946 |
true,
|
| 1947 |
false,
|
| 1948 |
false,
|
|
@@ -2053,7 +2053,7 @@
|
|
| 2053 |
],
|
| 2054 |
"model.decoder.layers.21.fc2.lora_E": [
|
| 2055 |
true,
|
| 2056 |
-
|
| 2057 |
true,
|
| 2058 |
true,
|
| 2059 |
false,
|
|
@@ -2067,12 +2067,12 @@
|
|
| 2067 |
],
|
| 2068 |
"model.decoder.layers.21.self_attn.k_proj.lora_E": [
|
| 2069 |
true,
|
| 2070 |
-
|
| 2071 |
-
|
| 2072 |
true,
|
| 2073 |
true,
|
| 2074 |
false,
|
| 2075 |
-
|
| 2076 |
false,
|
| 2077 |
true,
|
| 2078 |
true,
|
|
@@ -2094,17 +2094,17 @@
|
|
| 2094 |
false
|
| 2095 |
],
|
| 2096 |
"model.decoder.layers.21.self_attn.q_proj.lora_E": [
|
| 2097 |
-
true,
|
| 2098 |
false,
|
| 2099 |
false,
|
| 2100 |
false,
|
| 2101 |
-
true,
|
| 2102 |
-
true,
|
| 2103 |
false,
|
| 2104 |
false,
|
| 2105 |
true,
|
|
|
|
|
|
|
| 2106 |
true,
|
| 2107 |
false,
|
|
|
|
| 2108 |
false
|
| 2109 |
],
|
| 2110 |
"model.decoder.layers.21.self_attn.v_proj.lora_E": [
|
|
@@ -2217,7 +2217,7 @@
|
|
| 2217 |
false,
|
| 2218 |
false,
|
| 2219 |
false,
|
| 2220 |
-
|
| 2221 |
],
|
| 2222 |
"model.decoder.layers.22.self_attn.out_proj.lora_E": [
|
| 2223 |
false,
|
|
@@ -2226,7 +2226,7 @@
|
|
| 2226 |
true,
|
| 2227 |
false,
|
| 2228 |
true,
|
| 2229 |
-
|
| 2230 |
true,
|
| 2231 |
true,
|
| 2232 |
false,
|
|
@@ -2257,7 +2257,7 @@
|
|
| 2257 |
true,
|
| 2258 |
true,
|
| 2259 |
true,
|
| 2260 |
-
|
| 2261 |
false,
|
| 2262 |
true
|
| 2263 |
],
|
|
@@ -2439,7 +2439,7 @@
|
|
| 2439 |
false,
|
| 2440 |
false,
|
| 2441 |
false,
|
| 2442 |
-
|
| 2443 |
false,
|
| 2444 |
false
|
| 2445 |
],
|
|
@@ -2481,7 +2481,7 @@
|
|
| 2481 |
true,
|
| 2482 |
true,
|
| 2483 |
true,
|
| 2484 |
-
|
| 2485 |
false,
|
| 2486 |
true
|
| 2487 |
],
|
|
@@ -2492,7 +2492,7 @@
|
|
| 2492 |
true,
|
| 2493 |
true,
|
| 2494 |
true,
|
| 2495 |
-
|
| 2496 |
true,
|
| 2497 |
true,
|
| 2498 |
true,
|
|
@@ -2508,7 +2508,7 @@
|
|
| 2508 |
true,
|
| 2509 |
false,
|
| 2510 |
true,
|
| 2511 |
-
|
| 2512 |
false,
|
| 2513 |
false,
|
| 2514 |
false
|
|
@@ -2536,7 +2536,7 @@
|
|
| 2536 |
true,
|
| 2537 |
true,
|
| 2538 |
false,
|
| 2539 |
-
|
| 2540 |
false,
|
| 2541 |
true,
|
| 2542 |
true
|
|
@@ -2558,10 +2558,10 @@
|
|
| 2558 |
"model.decoder.layers.25.encoder_attn.out_proj.lora_E": [
|
| 2559 |
true,
|
| 2560 |
true,
|
| 2561 |
-
|
| 2562 |
-
true,
|
| 2563 |
true,
|
| 2564 |
false,
|
|
|
|
| 2565 |
true,
|
| 2566 |
true,
|
| 2567 |
true,
|
|
@@ -2627,7 +2627,7 @@
|
|
| 2627 |
],
|
| 2628 |
"model.decoder.layers.25.self_attn.k_proj.lora_E": [
|
| 2629 |
true,
|
| 2630 |
-
|
| 2631 |
true,
|
| 2632 |
true,
|
| 2633 |
false,
|
|
@@ -2649,7 +2649,7 @@
|
|
| 2649 |
false,
|
| 2650 |
false,
|
| 2651 |
true,
|
| 2652 |
-
|
| 2653 |
false,
|
| 2654 |
false
|
| 2655 |
],
|
|
@@ -2698,7 +2698,7 @@
|
|
| 2698 |
"model.decoder.layers.26.encoder_attn.out_proj.lora_E": [
|
| 2699 |
true,
|
| 2700 |
true,
|
| 2701 |
-
|
| 2702 |
true,
|
| 2703 |
true,
|
| 2704 |
true,
|
|
@@ -2706,7 +2706,7 @@
|
|
| 2706 |
true,
|
| 2707 |
true,
|
| 2708 |
true,
|
| 2709 |
-
|
| 2710 |
true
|
| 2711 |
],
|
| 2712 |
"model.decoder.layers.26.encoder_attn.q_proj.lora_E": [
|
|
@@ -2768,7 +2768,7 @@
|
|
| 2768 |
"model.decoder.layers.26.self_attn.k_proj.lora_E": [
|
| 2769 |
false,
|
| 2770 |
true,
|
| 2771 |
-
|
| 2772 |
true,
|
| 2773 |
false,
|
| 2774 |
false,
|
|
@@ -2796,7 +2796,6 @@
|
|
| 2796 |
"model.decoder.layers.26.self_attn.q_proj.lora_E": [
|
| 2797 |
false,
|
| 2798 |
false,
|
| 2799 |
-
true,
|
| 2800 |
false,
|
| 2801 |
false,
|
| 2802 |
false,
|
|
@@ -2805,7 +2804,8 @@
|
|
| 2805 |
false,
|
| 2806 |
false,
|
| 2807 |
false,
|
| 2808 |
-
|
|
|
|
| 2809 |
],
|
| 2810 |
"model.decoder.layers.26.self_attn.v_proj.lora_E": [
|
| 2811 |
false,
|
|
@@ -2841,12 +2841,12 @@
|
|
| 2841 |
true,
|
| 2842 |
true,
|
| 2843 |
true,
|
|
|
|
| 2844 |
true,
|
| 2845 |
true,
|
| 2846 |
true,
|
| 2847 |
-
|
| 2848 |
-
|
| 2849 |
-
true,
|
| 2850 |
true
|
| 2851 |
],
|
| 2852 |
"model.decoder.layers.27.encoder_attn.q_proj.lora_E": [
|
|
@@ -2925,7 +2925,7 @@
|
|
| 2925 |
true,
|
| 2926 |
false,
|
| 2927 |
false,
|
| 2928 |
-
|
| 2929 |
false,
|
| 2930 |
false,
|
| 2931 |
false,
|
|
@@ -2977,7 +2977,7 @@
|
|
| 2977 |
],
|
| 2978 |
"model.decoder.layers.28.encoder_attn.out_proj.lora_E": [
|
| 2979 |
true,
|
| 2980 |
-
|
| 2981 |
true,
|
| 2982 |
true,
|
| 2983 |
true,
|
|
@@ -3116,18 +3116,18 @@
|
|
| 3116 |
false
|
| 3117 |
],
|
| 3118 |
"model.decoder.layers.29.encoder_attn.out_proj.lora_E": [
|
| 3119 |
-
true,
|
| 3120 |
-
true,
|
| 3121 |
true,
|
| 3122 |
false,
|
| 3123 |
false,
|
| 3124 |
false,
|
| 3125 |
-
true,
|
| 3126 |
false,
|
| 3127 |
false,
|
| 3128 |
false,
|
| 3129 |
false,
|
| 3130 |
-
|
|
|
|
|
|
|
|
|
|
| 3131 |
],
|
| 3132 |
"model.decoder.layers.29.encoder_attn.q_proj.lora_E": [
|
| 3133 |
false,
|
|
@@ -3259,9 +3259,9 @@
|
|
| 3259 |
true,
|
| 3260 |
true,
|
| 3261 |
false,
|
|
|
|
| 3262 |
true,
|
| 3263 |
-
|
| 3264 |
-
true,
|
| 3265 |
true,
|
| 3266 |
true,
|
| 3267 |
false,
|
|
@@ -3396,17 +3396,17 @@
|
|
| 3396 |
false
|
| 3397 |
],
|
| 3398 |
"model.decoder.layers.30.encoder_attn.out_proj.lora_E": [
|
| 3399 |
-
true,
|
| 3400 |
true,
|
| 3401 |
false,
|
| 3402 |
false,
|
| 3403 |
false,
|
| 3404 |
false,
|
| 3405 |
-
|
|
|
|
|
|
|
| 3406 |
false,
|
| 3407 |
true,
|
| 3408 |
-
|
| 3409 |
-
true,
|
| 3410 |
false
|
| 3411 |
],
|
| 3412 |
"model.decoder.layers.30.encoder_attn.q_proj.lora_E": [
|
|
@@ -3509,7 +3509,7 @@
|
|
| 3509 |
],
|
| 3510 |
"model.decoder.layers.30.self_attn.v_proj.lora_E": [
|
| 3511 |
false,
|
| 3512 |
-
|
| 3513 |
true,
|
| 3514 |
false,
|
| 3515 |
false,
|
|
@@ -3517,7 +3517,7 @@
|
|
| 3517 |
false,
|
| 3518 |
false,
|
| 3519 |
false,
|
| 3520 |
-
|
| 3521 |
false,
|
| 3522 |
true
|
| 3523 |
],
|
|
@@ -3566,7 +3566,7 @@
|
|
| 3566 |
"model.decoder.layers.31.encoder_attn.v_proj.lora_E": [
|
| 3567 |
false,
|
| 3568 |
false,
|
| 3569 |
-
|
| 3570 |
false,
|
| 3571 |
true,
|
| 3572 |
false,
|
|
@@ -3622,7 +3622,7 @@
|
|
| 3622 |
"model.decoder.layers.31.self_attn.out_proj.lora_E": [
|
| 3623 |
false,
|
| 3624 |
false,
|
| 3625 |
-
|
| 3626 |
false,
|
| 3627 |
false,
|
| 3628 |
false,
|
|
@@ -3704,18 +3704,18 @@
|
|
| 3704 |
false
|
| 3705 |
],
|
| 3706 |
"model.decoder.layers.4.encoder_attn.v_proj.lora_E": [
|
| 3707 |
-
true,
|
| 3708 |
true,
|
| 3709 |
false,
|
| 3710 |
-
|
|
|
|
| 3711 |
false,
|
| 3712 |
false,
|
| 3713 |
true,
|
| 3714 |
true,
|
| 3715 |
false,
|
| 3716 |
true,
|
| 3717 |
-
|
| 3718 |
-
|
| 3719 |
],
|
| 3720 |
"model.decoder.layers.4.fc1.lora_E": [
|
| 3721 |
true,
|
|
@@ -3853,7 +3853,7 @@
|
|
| 3853 |
false,
|
| 3854 |
true,
|
| 3855 |
true,
|
| 3856 |
-
|
| 3857 |
false,
|
| 3858 |
true
|
| 3859 |
],
|
|
@@ -4331,7 +4331,7 @@
|
|
| 4331 |
false,
|
| 4332 |
false,
|
| 4333 |
false,
|
| 4334 |
-
|
| 4335 |
],
|
| 4336 |
"model.decoder.layers.8.self_attn.q_proj.lora_E": [
|
| 4337 |
false,
|
|
@@ -4443,7 +4443,7 @@
|
|
| 4443 |
true,
|
| 4444 |
true,
|
| 4445 |
true,
|
| 4446 |
-
|
| 4447 |
],
|
| 4448 |
"model.decoder.layers.9.self_attn.k_proj.lora_E": [
|
| 4449 |
false,
|
|
@@ -4550,8 +4550,8 @@
|
|
| 4550 |
true,
|
| 4551 |
false,
|
| 4552 |
false,
|
| 4553 |
-
|
| 4554 |
-
|
| 4555 |
false,
|
| 4556 |
false,
|
| 4557 |
false,
|
|
@@ -5015,7 +5015,7 @@
|
|
| 5015 |
false,
|
| 5016 |
true,
|
| 5017 |
false,
|
| 5018 |
-
|
| 5019 |
false,
|
| 5020 |
true
|
| 5021 |
],
|
|
@@ -5136,7 +5136,7 @@
|
|
| 5136 |
true,
|
| 5137 |
false,
|
| 5138 |
false,
|
| 5139 |
-
|
| 5140 |
true,
|
| 5141 |
false,
|
| 5142 |
false,
|
|
@@ -5223,7 +5223,7 @@
|
|
| 5223 |
true,
|
| 5224 |
false,
|
| 5225 |
true,
|
| 5226 |
-
|
| 5227 |
false,
|
| 5228 |
false,
|
| 5229 |
true,
|
|
@@ -5307,7 +5307,7 @@
|
|
| 5307 |
true,
|
| 5308 |
true,
|
| 5309 |
false,
|
| 5310 |
-
|
| 5311 |
true,
|
| 5312 |
false,
|
| 5313 |
true,
|
|
@@ -5384,16 +5384,16 @@
|
|
| 5384 |
false
|
| 5385 |
],
|
| 5386 |
"model.encoder.layers.18.self_attn.out_proj.lora_E": [
|
| 5387 |
-
true,
|
| 5388 |
false,
|
| 5389 |
-
|
|
|
|
|
|
|
| 5390 |
false,
|
| 5391 |
false,
|
| 5392 |
false,
|
| 5393 |
false,
|
| 5394 |
true,
|
| 5395 |
-
|
| 5396 |
-
true,
|
| 5397 |
false,
|
| 5398 |
false
|
| 5399 |
],
|
|
@@ -5806,8 +5806,8 @@
|
|
| 5806 |
"model.encoder.layers.22.self_attn.out_proj.lora_E": [
|
| 5807 |
false,
|
| 5808 |
false,
|
| 5809 |
-
|
| 5810 |
-
|
| 5811 |
false,
|
| 5812 |
true,
|
| 5813 |
false,
|
|
@@ -5898,7 +5898,7 @@
|
|
| 5898 |
true,
|
| 5899 |
false,
|
| 5900 |
false,
|
| 5901 |
-
|
| 5902 |
true
|
| 5903 |
],
|
| 5904 |
"model.encoder.layers.23.self_attn.q_proj.lora_E": [
|
|
@@ -6063,7 +6063,7 @@
|
|
| 6063 |
false,
|
| 6064 |
false,
|
| 6065 |
true,
|
| 6066 |
-
|
| 6067 |
false,
|
| 6068 |
false,
|
| 6069 |
false,
|
|
@@ -6425,7 +6425,7 @@
|
|
| 6425 |
true,
|
| 6426 |
true,
|
| 6427 |
true,
|
| 6428 |
-
|
| 6429 |
true,
|
| 6430 |
true,
|
| 6431 |
true,
|
|
@@ -6569,7 +6569,7 @@
|
|
| 6569 |
true,
|
| 6570 |
true,
|
| 6571 |
true,
|
| 6572 |
-
|
| 6573 |
true,
|
| 6574 |
false
|
| 6575 |
],
|
|
@@ -6677,8 +6677,8 @@
|
|
| 6677 |
true,
|
| 6678 |
true,
|
| 6679 |
false,
|
| 6680 |
-
|
| 6681 |
-
|
| 6682 |
true,
|
| 6683 |
true,
|
| 6684 |
true,
|
|
@@ -6794,7 +6794,7 @@
|
|
| 6794 |
false,
|
| 6795 |
false,
|
| 6796 |
false,
|
| 6797 |
-
|
| 6798 |
false
|
| 6799 |
],
|
| 6800 |
"model.encoder.layers.5.self_attn.k_proj.lora_E": [
|
|
@@ -7182,7 +7182,7 @@
|
|
| 7182 |
false,
|
| 7183 |
false,
|
| 7184 |
false,
|
| 7185 |
-
|
| 7186 |
false,
|
| 7187 |
false,
|
| 7188 |
false,
|
|
|
|
| 210 |
true,
|
| 211 |
false,
|
| 212 |
false,
|
|
|
|
| 213 |
false,
|
| 214 |
false,
|
| 215 |
false,
|
| 216 |
false,
|
| 217 |
+
false,
|
| 218 |
+
false
|
| 219 |
],
|
| 220 |
"model.decoder.layers.1.fc1.lora_E": [
|
| 221 |
true,
|
|
|
|
| 365 |
true,
|
| 366 |
true,
|
| 367 |
true,
|
| 368 |
+
false,
|
| 369 |
true,
|
| 370 |
true,
|
| 371 |
true,
|
|
|
|
| 549 |
false,
|
| 550 |
false,
|
| 551 |
false,
|
| 552 |
+
false,
|
| 553 |
true,
|
| 554 |
true
|
| 555 |
],
|
|
|
|
| 574 |
false,
|
| 575 |
false,
|
| 576 |
false,
|
| 577 |
+
false,
|
| 578 |
false,
|
| 579 |
false,
|
| 580 |
false,
|
|
|
|
| 656 |
true,
|
| 657 |
true,
|
| 658 |
true,
|
| 659 |
+
false,
|
| 660 |
true,
|
| 661 |
true,
|
| 662 |
true,
|
|
|
|
| 1156 |
false
|
| 1157 |
],
|
| 1158 |
"model.decoder.layers.16.encoder_attn.out_proj.lora_E": [
|
| 1159 |
+
false,
|
| 1160 |
true,
|
| 1161 |
false,
|
| 1162 |
false,
|
|
|
|
| 1296 |
false
|
| 1297 |
],
|
| 1298 |
"model.decoder.layers.17.encoder_attn.out_proj.lora_E": [
|
| 1299 |
+
false,
|
| 1300 |
false,
|
| 1301 |
true,
|
| 1302 |
false,
|
|
|
|
| 1306 |
false,
|
| 1307 |
false,
|
| 1308 |
false,
|
| 1309 |
+
false,
|
| 1310 |
false
|
| 1311 |
],
|
| 1312 |
"model.decoder.layers.17.encoder_attn.q_proj.lora_E": [
|
|
|
|
| 1461 |
false,
|
| 1462 |
false,
|
| 1463 |
false,
|
| 1464 |
+
false
|
| 1465 |
],
|
| 1466 |
"model.decoder.layers.18.encoder_attn.v_proj.lora_E": [
|
| 1467 |
true,
|
|
|
|
| 1521 |
],
|
| 1522 |
"model.decoder.layers.18.self_attn.out_proj.lora_E": [
|
| 1523 |
true,
|
| 1524 |
+
false,
|
| 1525 |
true,
|
| 1526 |
false,
|
| 1527 |
true,
|
|
|
|
| 1529 |
true,
|
| 1530 |
false,
|
| 1531 |
true,
|
| 1532 |
+
false,
|
| 1533 |
true,
|
| 1534 |
true
|
| 1535 |
],
|
|
|
|
| 1577 |
],
|
| 1578 |
"model.decoder.layers.19.encoder_attn.out_proj.lora_E": [
|
| 1579 |
false,
|
|
|
|
| 1580 |
false,
|
| 1581 |
false,
|
| 1582 |
false,
|
| 1583 |
false,
|
| 1584 |
false,
|
| 1585 |
+
false,
|
| 1586 |
+
false,
|
| 1587 |
false,
|
| 1588 |
false,
|
| 1589 |
false,
|
|
|
|
| 1668 |
true,
|
| 1669 |
false,
|
| 1670 |
false,
|
| 1671 |
+
false,
|
| 1672 |
true,
|
| 1673 |
false,
|
| 1674 |
false
|
|
|
|
| 1699 |
false,
|
| 1700 |
true,
|
| 1701 |
false,
|
| 1702 |
+
false
|
| 1703 |
],
|
| 1704 |
"model.decoder.layers.2.encoder_attn.k_proj.lora_E": [
|
| 1705 |
false,
|
|
|
|
| 1800 |
false
|
| 1801 |
],
|
| 1802 |
"model.decoder.layers.2.self_attn.out_proj.lora_E": [
|
|
|
|
| 1803 |
true,
|
| 1804 |
false,
|
| 1805 |
false,
|
|
|
|
|
|
|
| 1806 |
false,
|
| 1807 |
true,
|
| 1808 |
+
false,
|
| 1809 |
+
false,
|
| 1810 |
+
false,
|
| 1811 |
true,
|
| 1812 |
true,
|
| 1813 |
true,
|
| 1814 |
+
false
|
| 1815 |
],
|
| 1816 |
"model.decoder.layers.2.self_attn.q_proj.lora_E": [
|
| 1817 |
false,
|
|
|
|
| 1863 |
true,
|
| 1864 |
false,
|
| 1865 |
true,
|
| 1866 |
+
false,
|
| 1867 |
true,
|
| 1868 |
true,
|
| 1869 |
true,
|
|
|
|
| 1922 |
true,
|
| 1923 |
true,
|
| 1924 |
true,
|
| 1925 |
+
false,
|
| 1926 |
true
|
| 1927 |
],
|
| 1928 |
"model.decoder.layers.20.self_attn.k_proj.lora_E": [
|
|
|
|
| 1942 |
"model.decoder.layers.20.self_attn.out_proj.lora_E": [
|
| 1943 |
true,
|
| 1944 |
false,
|
| 1945 |
+
false,
|
| 1946 |
true,
|
| 1947 |
false,
|
| 1948 |
false,
|
|
|
|
| 2053 |
],
|
| 2054 |
"model.decoder.layers.21.fc2.lora_E": [
|
| 2055 |
true,
|
| 2056 |
+
false,
|
| 2057 |
true,
|
| 2058 |
true,
|
| 2059 |
false,
|
|
|
|
| 2067 |
],
|
| 2068 |
"model.decoder.layers.21.self_attn.k_proj.lora_E": [
|
| 2069 |
true,
|
| 2070 |
+
false,
|
| 2071 |
+
false,
|
| 2072 |
true,
|
| 2073 |
true,
|
| 2074 |
false,
|
| 2075 |
+
false,
|
| 2076 |
false,
|
| 2077 |
true,
|
| 2078 |
true,
|
|
|
|
| 2094 |
false
|
| 2095 |
],
|
| 2096 |
"model.decoder.layers.21.self_attn.q_proj.lora_E": [
|
|
|
|
| 2097 |
false,
|
| 2098 |
false,
|
| 2099 |
false,
|
|
|
|
|
|
|
| 2100 |
false,
|
| 2101 |
false,
|
| 2102 |
true,
|
| 2103 |
+
false,
|
| 2104 |
+
false,
|
| 2105 |
true,
|
| 2106 |
false,
|
| 2107 |
+
false,
|
| 2108 |
false
|
| 2109 |
],
|
| 2110 |
"model.decoder.layers.21.self_attn.v_proj.lora_E": [
|
|
|
|
| 2217 |
false,
|
| 2218 |
false,
|
| 2219 |
false,
|
| 2220 |
+
false
|
| 2221 |
],
|
| 2222 |
"model.decoder.layers.22.self_attn.out_proj.lora_E": [
|
| 2223 |
false,
|
|
|
|
| 2226 |
true,
|
| 2227 |
false,
|
| 2228 |
true,
|
| 2229 |
+
false,
|
| 2230 |
true,
|
| 2231 |
true,
|
| 2232 |
false,
|
|
|
|
| 2257 |
true,
|
| 2258 |
true,
|
| 2259 |
true,
|
| 2260 |
+
false,
|
| 2261 |
false,
|
| 2262 |
true
|
| 2263 |
],
|
|
|
|
| 2439 |
false,
|
| 2440 |
false,
|
| 2441 |
false,
|
| 2442 |
+
false,
|
| 2443 |
false,
|
| 2444 |
false
|
| 2445 |
],
|
|
|
|
| 2481 |
true,
|
| 2482 |
true,
|
| 2483 |
true,
|
| 2484 |
+
false,
|
| 2485 |
false,
|
| 2486 |
true
|
| 2487 |
],
|
|
|
|
| 2492 |
true,
|
| 2493 |
true,
|
| 2494 |
true,
|
| 2495 |
+
false,
|
| 2496 |
true,
|
| 2497 |
true,
|
| 2498 |
true,
|
|
|
|
| 2508 |
true,
|
| 2509 |
false,
|
| 2510 |
true,
|
| 2511 |
+
false,
|
| 2512 |
false,
|
| 2513 |
false,
|
| 2514 |
false
|
|
|
|
| 2536 |
true,
|
| 2537 |
true,
|
| 2538 |
false,
|
| 2539 |
+
false,
|
| 2540 |
false,
|
| 2541 |
true,
|
| 2542 |
true
|
|
|
|
| 2558 |
"model.decoder.layers.25.encoder_attn.out_proj.lora_E": [
|
| 2559 |
true,
|
| 2560 |
true,
|
| 2561 |
+
false,
|
|
|
|
| 2562 |
true,
|
| 2563 |
false,
|
| 2564 |
+
false,
|
| 2565 |
true,
|
| 2566 |
true,
|
| 2567 |
true,
|
|
|
|
| 2627 |
],
|
| 2628 |
"model.decoder.layers.25.self_attn.k_proj.lora_E": [
|
| 2629 |
true,
|
| 2630 |
+
false,
|
| 2631 |
true,
|
| 2632 |
true,
|
| 2633 |
false,
|
|
|
|
| 2649 |
false,
|
| 2650 |
false,
|
| 2651 |
true,
|
| 2652 |
+
false,
|
| 2653 |
false,
|
| 2654 |
false
|
| 2655 |
],
|
|
|
|
| 2698 |
"model.decoder.layers.26.encoder_attn.out_proj.lora_E": [
|
| 2699 |
true,
|
| 2700 |
true,
|
| 2701 |
+
false,
|
| 2702 |
true,
|
| 2703 |
true,
|
| 2704 |
true,
|
|
|
|
| 2706 |
true,
|
| 2707 |
true,
|
| 2708 |
true,
|
| 2709 |
+
false,
|
| 2710 |
true
|
| 2711 |
],
|
| 2712 |
"model.decoder.layers.26.encoder_attn.q_proj.lora_E": [
|
|
|
|
| 2768 |
"model.decoder.layers.26.self_attn.k_proj.lora_E": [
|
| 2769 |
false,
|
| 2770 |
true,
|
| 2771 |
+
false,
|
| 2772 |
true,
|
| 2773 |
false,
|
| 2774 |
false,
|
|
|
|
| 2796 |
"model.decoder.layers.26.self_attn.q_proj.lora_E": [
|
| 2797 |
false,
|
| 2798 |
false,
|
|
|
|
| 2799 |
false,
|
| 2800 |
false,
|
| 2801 |
false,
|
|
|
|
| 2804 |
false,
|
| 2805 |
false,
|
| 2806 |
false,
|
| 2807 |
+
false,
|
| 2808 |
+
false
|
| 2809 |
],
|
| 2810 |
"model.decoder.layers.26.self_attn.v_proj.lora_E": [
|
| 2811 |
false,
|
|
|
|
| 2841 |
true,
|
| 2842 |
true,
|
| 2843 |
true,
|
| 2844 |
+
false,
|
| 2845 |
true,
|
| 2846 |
true,
|
| 2847 |
true,
|
| 2848 |
+
false,
|
| 2849 |
+
false,
|
|
|
|
| 2850 |
true
|
| 2851 |
],
|
| 2852 |
"model.decoder.layers.27.encoder_attn.q_proj.lora_E": [
|
|
|
|
| 2925 |
true,
|
| 2926 |
false,
|
| 2927 |
false,
|
| 2928 |
+
false,
|
| 2929 |
false,
|
| 2930 |
false,
|
| 2931 |
false,
|
|
|
|
| 2977 |
],
|
| 2978 |
"model.decoder.layers.28.encoder_attn.out_proj.lora_E": [
|
| 2979 |
true,
|
| 2980 |
+
false,
|
| 2981 |
true,
|
| 2982 |
true,
|
| 2983 |
true,
|
|
|
|
| 3116 |
false
|
| 3117 |
],
|
| 3118 |
"model.decoder.layers.29.encoder_attn.out_proj.lora_E": [
|
|
|
|
|
|
|
| 3119 |
true,
|
| 3120 |
false,
|
| 3121 |
false,
|
| 3122 |
false,
|
|
|
|
| 3123 |
false,
|
| 3124 |
false,
|
| 3125 |
false,
|
| 3126 |
false,
|
| 3127 |
+
false,
|
| 3128 |
+
false,
|
| 3129 |
+
false,
|
| 3130 |
+
false
|
| 3131 |
],
|
| 3132 |
"model.decoder.layers.29.encoder_attn.q_proj.lora_E": [
|
| 3133 |
false,
|
|
|
|
| 3259 |
true,
|
| 3260 |
true,
|
| 3261 |
false,
|
| 3262 |
+
false,
|
| 3263 |
true,
|
| 3264 |
+
false,
|
|
|
|
| 3265 |
true,
|
| 3266 |
true,
|
| 3267 |
false,
|
|
|
|
| 3396 |
false
|
| 3397 |
],
|
| 3398 |
"model.decoder.layers.30.encoder_attn.out_proj.lora_E": [
|
|
|
|
| 3399 |
true,
|
| 3400 |
false,
|
| 3401 |
false,
|
| 3402 |
false,
|
| 3403 |
false,
|
| 3404 |
+
false,
|
| 3405 |
+
false,
|
| 3406 |
+
false,
|
| 3407 |
false,
|
| 3408 |
true,
|
| 3409 |
+
false,
|
|
|
|
| 3410 |
false
|
| 3411 |
],
|
| 3412 |
"model.decoder.layers.30.encoder_attn.q_proj.lora_E": [
|
|
|
|
| 3509 |
],
|
| 3510 |
"model.decoder.layers.30.self_attn.v_proj.lora_E": [
|
| 3511 |
false,
|
| 3512 |
+
false,
|
| 3513 |
true,
|
| 3514 |
false,
|
| 3515 |
false,
|
|
|
|
| 3517 |
false,
|
| 3518 |
false,
|
| 3519 |
false,
|
| 3520 |
+
false,
|
| 3521 |
false,
|
| 3522 |
true
|
| 3523 |
],
|
|
|
|
| 3566 |
"model.decoder.layers.31.encoder_attn.v_proj.lora_E": [
|
| 3567 |
false,
|
| 3568 |
false,
|
| 3569 |
+
false,
|
| 3570 |
false,
|
| 3571 |
true,
|
| 3572 |
false,
|
|
|
|
| 3622 |
"model.decoder.layers.31.self_attn.out_proj.lora_E": [
|
| 3623 |
false,
|
| 3624 |
false,
|
| 3625 |
+
false,
|
| 3626 |
false,
|
| 3627 |
false,
|
| 3628 |
false,
|
|
|
|
| 3704 |
false
|
| 3705 |
],
|
| 3706 |
"model.decoder.layers.4.encoder_attn.v_proj.lora_E": [
|
|
|
|
| 3707 |
true,
|
| 3708 |
false,
|
| 3709 |
+
false,
|
| 3710 |
+
false,
|
| 3711 |
false,
|
| 3712 |
false,
|
| 3713 |
true,
|
| 3714 |
true,
|
| 3715 |
false,
|
| 3716 |
true,
|
| 3717 |
+
false,
|
| 3718 |
+
false
|
| 3719 |
],
|
| 3720 |
"model.decoder.layers.4.fc1.lora_E": [
|
| 3721 |
true,
|
|
|
|
| 3853 |
false,
|
| 3854 |
true,
|
| 3855 |
true,
|
| 3856 |
+
false,
|
| 3857 |
false,
|
| 3858 |
true
|
| 3859 |
],
|
|
|
|
| 4331 |
false,
|
| 4332 |
false,
|
| 4333 |
false,
|
| 4334 |
+
false
|
| 4335 |
],
|
| 4336 |
"model.decoder.layers.8.self_attn.q_proj.lora_E": [
|
| 4337 |
false,
|
|
|
|
| 4443 |
true,
|
| 4444 |
true,
|
| 4445 |
true,
|
| 4446 |
+
false
|
| 4447 |
],
|
| 4448 |
"model.decoder.layers.9.self_attn.k_proj.lora_E": [
|
| 4449 |
false,
|
|
|
|
| 4550 |
true,
|
| 4551 |
false,
|
| 4552 |
false,
|
| 4553 |
+
false,
|
| 4554 |
+
false,
|
| 4555 |
false,
|
| 4556 |
false,
|
| 4557 |
false,
|
|
|
|
| 5015 |
false,
|
| 5016 |
true,
|
| 5017 |
false,
|
| 5018 |
+
false,
|
| 5019 |
false,
|
| 5020 |
true
|
| 5021 |
],
|
|
|
|
| 5136 |
true,
|
| 5137 |
false,
|
| 5138 |
false,
|
| 5139 |
+
false,
|
| 5140 |
true,
|
| 5141 |
false,
|
| 5142 |
false,
|
|
|
|
| 5223 |
true,
|
| 5224 |
false,
|
| 5225 |
true,
|
| 5226 |
+
false,
|
| 5227 |
false,
|
| 5228 |
false,
|
| 5229 |
true,
|
|
|
|
| 5307 |
true,
|
| 5308 |
true,
|
| 5309 |
false,
|
| 5310 |
+
false,
|
| 5311 |
true,
|
| 5312 |
false,
|
| 5313 |
true,
|
|
|
|
| 5384 |
false
|
| 5385 |
],
|
| 5386 |
"model.encoder.layers.18.self_attn.out_proj.lora_E": [
|
|
|
|
| 5387 |
false,
|
| 5388 |
+
false,
|
| 5389 |
+
false,
|
| 5390 |
+
false,
|
| 5391 |
false,
|
| 5392 |
false,
|
| 5393 |
false,
|
| 5394 |
false,
|
| 5395 |
true,
|
| 5396 |
+
false,
|
|
|
|
| 5397 |
false,
|
| 5398 |
false
|
| 5399 |
],
|
|
|
|
| 5806 |
"model.encoder.layers.22.self_attn.out_proj.lora_E": [
|
| 5807 |
false,
|
| 5808 |
false,
|
| 5809 |
+
false,
|
| 5810 |
+
false,
|
| 5811 |
false,
|
| 5812 |
true,
|
| 5813 |
false,
|
|
|
|
| 5898 |
true,
|
| 5899 |
false,
|
| 5900 |
false,
|
| 5901 |
+
false,
|
| 5902 |
true
|
| 5903 |
],
|
| 5904 |
"model.encoder.layers.23.self_attn.q_proj.lora_E": [
|
|
|
|
| 6063 |
false,
|
| 6064 |
false,
|
| 6065 |
true,
|
| 6066 |
+
false,
|
| 6067 |
false,
|
| 6068 |
false,
|
| 6069 |
false,
|
|
|
|
| 6425 |
true,
|
| 6426 |
true,
|
| 6427 |
true,
|
| 6428 |
+
false,
|
| 6429 |
true,
|
| 6430 |
true,
|
| 6431 |
true,
|
|
|
|
| 6569 |
true,
|
| 6570 |
true,
|
| 6571 |
true,
|
| 6572 |
+
false,
|
| 6573 |
true,
|
| 6574 |
false
|
| 6575 |
],
|
|
|
|
| 6677 |
true,
|
| 6678 |
true,
|
| 6679 |
false,
|
| 6680 |
+
false,
|
| 6681 |
+
false,
|
| 6682 |
true,
|
| 6683 |
true,
|
| 6684 |
true,
|
|
|
|
| 6794 |
false,
|
| 6795 |
false,
|
| 6796 |
false,
|
| 6797 |
+
false,
|
| 6798 |
false
|
| 6799 |
],
|
| 6800 |
"model.encoder.layers.5.self_attn.k_proj.lora_E": [
|
|
|
|
| 7182 |
false,
|
| 7183 |
false,
|
| 7184 |
false,
|
| 7185 |
+
false,
|
| 7186 |
false,
|
| 7187 |
false,
|
| 7188 |
false,
|
adapter_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:067f61cbd367196dd55812c979bcc90a913f1a57880157f22943d127e4c40ecd
|
| 3 |
+
size 38697637
|
best_checkpoint/README.md
CHANGED
|
@@ -268,6 +268,90 @@ The following `bitsandbytes` quantization config was used during training:
|
|
| 268 |
- bnb_4bit_use_double_quant: False
|
| 269 |
- bnb_4bit_compute_dtype: float32
|
| 270 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 271 |
The following `bitsandbytes` quantization config was used during training:
|
| 272 |
- quant_method: bitsandbytes
|
| 273 |
- load_in_8bit: True
|
|
@@ -303,5 +387,12 @@ The following `bitsandbytes` quantization config was used during training:
|
|
| 303 |
- PEFT 0.5.0
|
| 304 |
- PEFT 0.5.0
|
| 305 |
- PEFT 0.5.0
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 306 |
|
| 307 |
- PEFT 0.5.0
|
|
|
|
| 268 |
- bnb_4bit_use_double_quant: False
|
| 269 |
- bnb_4bit_compute_dtype: float32
|
| 270 |
|
| 271 |
+
The following `bitsandbytes` quantization config was used during training:
|
| 272 |
+
- quant_method: bitsandbytes
|
| 273 |
+
- load_in_8bit: True
|
| 274 |
+
- load_in_4bit: False
|
| 275 |
+
- llm_int8_threshold: 6.0
|
| 276 |
+
- llm_int8_skip_modules: None
|
| 277 |
+
- llm_int8_enable_fp32_cpu_offload: False
|
| 278 |
+
- llm_int8_has_fp16_weight: False
|
| 279 |
+
- bnb_4bit_quant_type: fp4
|
| 280 |
+
- bnb_4bit_use_double_quant: False
|
| 281 |
+
- bnb_4bit_compute_dtype: float32
|
| 282 |
+
|
| 283 |
+
The following `bitsandbytes` quantization config was used during training:
|
| 284 |
+
- quant_method: bitsandbytes
|
| 285 |
+
- load_in_8bit: True
|
| 286 |
+
- load_in_4bit: False
|
| 287 |
+
- llm_int8_threshold: 6.0
|
| 288 |
+
- llm_int8_skip_modules: None
|
| 289 |
+
- llm_int8_enable_fp32_cpu_offload: False
|
| 290 |
+
- llm_int8_has_fp16_weight: False
|
| 291 |
+
- bnb_4bit_quant_type: fp4
|
| 292 |
+
- bnb_4bit_use_double_quant: False
|
| 293 |
+
- bnb_4bit_compute_dtype: float32
|
| 294 |
+
|
| 295 |
+
The following `bitsandbytes` quantization config was used during training:
|
| 296 |
+
- quant_method: bitsandbytes
|
| 297 |
+
- load_in_8bit: True
|
| 298 |
+
- load_in_4bit: False
|
| 299 |
+
- llm_int8_threshold: 6.0
|
| 300 |
+
- llm_int8_skip_modules: None
|
| 301 |
+
- llm_int8_enable_fp32_cpu_offload: False
|
| 302 |
+
- llm_int8_has_fp16_weight: False
|
| 303 |
+
- bnb_4bit_quant_type: fp4
|
| 304 |
+
- bnb_4bit_use_double_quant: False
|
| 305 |
+
- bnb_4bit_compute_dtype: float32
|
| 306 |
+
|
| 307 |
+
The following `bitsandbytes` quantization config was used during training:
|
| 308 |
+
- quant_method: bitsandbytes
|
| 309 |
+
- load_in_8bit: True
|
| 310 |
+
- load_in_4bit: False
|
| 311 |
+
- llm_int8_threshold: 6.0
|
| 312 |
+
- llm_int8_skip_modules: None
|
| 313 |
+
- llm_int8_enable_fp32_cpu_offload: False
|
| 314 |
+
- llm_int8_has_fp16_weight: False
|
| 315 |
+
- bnb_4bit_quant_type: fp4
|
| 316 |
+
- bnb_4bit_use_double_quant: False
|
| 317 |
+
- bnb_4bit_compute_dtype: float32
|
| 318 |
+
|
| 319 |
+
The following `bitsandbytes` quantization config was used during training:
|
| 320 |
+
- quant_method: bitsandbytes
|
| 321 |
+
- load_in_8bit: True
|
| 322 |
+
- load_in_4bit: False
|
| 323 |
+
- llm_int8_threshold: 6.0
|
| 324 |
+
- llm_int8_skip_modules: None
|
| 325 |
+
- llm_int8_enable_fp32_cpu_offload: False
|
| 326 |
+
- llm_int8_has_fp16_weight: False
|
| 327 |
+
- bnb_4bit_quant_type: fp4
|
| 328 |
+
- bnb_4bit_use_double_quant: False
|
| 329 |
+
- bnb_4bit_compute_dtype: float32
|
| 330 |
+
|
| 331 |
+
The following `bitsandbytes` quantization config was used during training:
|
| 332 |
+
- quant_method: bitsandbytes
|
| 333 |
+
- load_in_8bit: True
|
| 334 |
+
- load_in_4bit: False
|
| 335 |
+
- llm_int8_threshold: 6.0
|
| 336 |
+
- llm_int8_skip_modules: None
|
| 337 |
+
- llm_int8_enable_fp32_cpu_offload: False
|
| 338 |
+
- llm_int8_has_fp16_weight: False
|
| 339 |
+
- bnb_4bit_quant_type: fp4
|
| 340 |
+
- bnb_4bit_use_double_quant: False
|
| 341 |
+
- bnb_4bit_compute_dtype: float32
|
| 342 |
+
|
| 343 |
+
The following `bitsandbytes` quantization config was used during training:
|
| 344 |
+
- quant_method: bitsandbytes
|
| 345 |
+
- load_in_8bit: True
|
| 346 |
+
- load_in_4bit: False
|
| 347 |
+
- llm_int8_threshold: 6.0
|
| 348 |
+
- llm_int8_skip_modules: None
|
| 349 |
+
- llm_int8_enable_fp32_cpu_offload: False
|
| 350 |
+
- llm_int8_has_fp16_weight: False
|
| 351 |
+
- bnb_4bit_quant_type: fp4
|
| 352 |
+
- bnb_4bit_use_double_quant: False
|
| 353 |
+
- bnb_4bit_compute_dtype: float32
|
| 354 |
+
|
| 355 |
The following `bitsandbytes` quantization config was used during training:
|
| 356 |
- quant_method: bitsandbytes
|
| 357 |
- load_in_8bit: True
|
|
|
|
| 387 |
- PEFT 0.5.0
|
| 388 |
- PEFT 0.5.0
|
| 389 |
- PEFT 0.5.0
|
| 390 |
+
- PEFT 0.5.0
|
| 391 |
+
- PEFT 0.5.0
|
| 392 |
+
- PEFT 0.5.0
|
| 393 |
+
- PEFT 0.5.0
|
| 394 |
+
- PEFT 0.5.0
|
| 395 |
+
- PEFT 0.5.0
|
| 396 |
+
- PEFT 0.5.0
|
| 397 |
|
| 398 |
- PEFT 0.5.0
|
best_checkpoint/adapter_config.json
CHANGED
|
@@ -38,7 +38,7 @@
|
|
| 38 |
"model.decoder.layers.0.encoder_attn.out_proj.lora_E": [
|
| 39 |
true,
|
| 40 |
false,
|
| 41 |
-
|
| 42 |
false,
|
| 43 |
false,
|
| 44 |
true,
|
|
@@ -73,7 +73,7 @@
|
|
| 73 |
false,
|
| 74 |
false,
|
| 75 |
false,
|
| 76 |
-
|
| 77 |
true,
|
| 78 |
true
|
| 79 |
],
|
|
@@ -102,7 +102,7 @@
|
|
| 102 |
true,
|
| 103 |
true,
|
| 104 |
true,
|
| 105 |
-
|
| 106 |
true
|
| 107 |
],
|
| 108 |
"model.decoder.layers.0.self_attn.k_proj.lora_E": [
|
|
@@ -215,7 +215,7 @@
|
|
| 215 |
false,
|
| 216 |
false,
|
| 217 |
false,
|
| 218 |
-
|
| 219 |
],
|
| 220 |
"model.decoder.layers.1.fc1.lora_E": [
|
| 221 |
true,
|
|
@@ -365,7 +365,7 @@
|
|
| 365 |
true,
|
| 366 |
true,
|
| 367 |
true,
|
| 368 |
-
|
| 369 |
true,
|
| 370 |
true,
|
| 371 |
true,
|
|
@@ -504,12 +504,12 @@
|
|
| 504 |
true,
|
| 505 |
true,
|
| 506 |
true,
|
| 507 |
-
false,
|
| 508 |
true,
|
| 509 |
true,
|
| 510 |
true,
|
| 511 |
true,
|
| 512 |
-
true
|
|
|
|
| 513 |
],
|
| 514 |
"model.decoder.layers.11.fc2.lora_E": [
|
| 515 |
true,
|
|
@@ -550,7 +550,7 @@
|
|
| 550 |
false,
|
| 551 |
false,
|
| 552 |
false,
|
| 553 |
-
|
| 554 |
true
|
| 555 |
],
|
| 556 |
"model.decoder.layers.11.self_attn.q_proj.lora_E": [
|
|
@@ -574,7 +574,7 @@
|
|
| 574 |
false,
|
| 575 |
false,
|
| 576 |
false,
|
| 577 |
-
|
| 578 |
false,
|
| 579 |
false,
|
| 580 |
false,
|
|
@@ -680,17 +680,17 @@
|
|
| 680 |
false
|
| 681 |
],
|
| 682 |
"model.decoder.layers.12.self_attn.out_proj.lora_E": [
|
| 683 |
-
true,
|
| 684 |
false,
|
| 685 |
false,
|
| 686 |
-
true,
|
| 687 |
false,
|
| 688 |
false,
|
| 689 |
false,
|
| 690 |
false,
|
| 691 |
-
true,
|
| 692 |
false,
|
| 693 |
false,
|
|
|
|
|
|
|
|
|
|
| 694 |
false
|
| 695 |
],
|
| 696 |
"model.decoder.layers.12.self_attn.q_proj.lora_E": [
|
|
@@ -737,17 +737,17 @@
|
|
| 737 |
],
|
| 738 |
"model.decoder.layers.13.encoder_attn.out_proj.lora_E": [
|
| 739 |
false,
|
| 740 |
-
|
| 741 |
false,
|
| 742 |
true,
|
| 743 |
false,
|
| 744 |
false,
|
| 745 |
false,
|
| 746 |
false,
|
| 747 |
-
true,
|
| 748 |
false,
|
| 749 |
-
|
| 750 |
-
|
|
|
|
| 751 |
],
|
| 752 |
"model.decoder.layers.13.encoder_attn.q_proj.lora_E": [
|
| 753 |
false,
|
|
@@ -765,14 +765,14 @@
|
|
| 765 |
],
|
| 766 |
"model.decoder.layers.13.encoder_attn.v_proj.lora_E": [
|
| 767 |
false,
|
| 768 |
-
|
| 769 |
false,
|
| 770 |
false,
|
| 771 |
true,
|
| 772 |
true,
|
| 773 |
true,
|
| 774 |
-
|
| 775 |
-
|
| 776 |
true,
|
| 777 |
true,
|
| 778 |
true
|
|
@@ -823,7 +823,7 @@
|
|
| 823 |
false,
|
| 824 |
false,
|
| 825 |
true,
|
| 826 |
-
|
| 827 |
true,
|
| 828 |
false,
|
| 829 |
true,
|
|
@@ -859,7 +859,7 @@
|
|
| 859 |
false,
|
| 860 |
false,
|
| 861 |
false,
|
| 862 |
-
|
| 863 |
],
|
| 864 |
"model.decoder.layers.14.encoder_attn.k_proj.lora_E": [
|
| 865 |
false,
|
|
@@ -927,7 +927,7 @@
|
|
| 927 |
true,
|
| 928 |
true,
|
| 929 |
true,
|
| 930 |
-
|
| 931 |
true,
|
| 932 |
true
|
| 933 |
],
|
|
@@ -962,7 +962,7 @@
|
|
| 962 |
"model.decoder.layers.14.self_attn.out_proj.lora_E": [
|
| 963 |
false,
|
| 964 |
true,
|
| 965 |
-
|
| 966 |
true,
|
| 967 |
false,
|
| 968 |
false,
|
|
@@ -1077,7 +1077,7 @@
|
|
| 1077 |
true,
|
| 1078 |
true,
|
| 1079 |
true,
|
| 1080 |
-
|
| 1081 |
true,
|
| 1082 |
true,
|
| 1083 |
true,
|
|
@@ -1108,7 +1108,7 @@
|
|
| 1108 |
true,
|
| 1109 |
true,
|
| 1110 |
false,
|
| 1111 |
-
|
| 1112 |
false,
|
| 1113 |
false,
|
| 1114 |
true
|
|
@@ -1128,17 +1128,17 @@
|
|
| 1128 |
false
|
| 1129 |
],
|
| 1130 |
"model.decoder.layers.15.self_attn.v_proj.lora_E": [
|
| 1131 |
-
true,
|
| 1132 |
-
true,
|
| 1133 |
-
true,
|
| 1134 |
-
true,
|
| 1135 |
true,
|
| 1136 |
false,
|
|
|
|
| 1137 |
false,
|
|
|
|
| 1138 |
false,
|
| 1139 |
false,
|
| 1140 |
true,
|
| 1141 |
true,
|
|
|
|
|
|
|
| 1142 |
false
|
| 1143 |
],
|
| 1144 |
"model.decoder.layers.16.encoder_attn.k_proj.lora_E": [
|
|
@@ -1181,7 +1181,7 @@
|
|
| 1181 |
false,
|
| 1182 |
false,
|
| 1183 |
false,
|
| 1184 |
-
|
| 1185 |
],
|
| 1186 |
"model.decoder.layers.16.encoder_attn.v_proj.lora_E": [
|
| 1187 |
true,
|
|
@@ -1221,7 +1221,7 @@
|
|
| 1221 |
true,
|
| 1222 |
true,
|
| 1223 |
true,
|
| 1224 |
-
|
| 1225 |
true,
|
| 1226 |
true
|
| 1227 |
],
|
|
@@ -1272,7 +1272,7 @@
|
|
| 1272 |
true,
|
| 1273 |
true,
|
| 1274 |
true,
|
| 1275 |
-
|
| 1276 |
true,
|
| 1277 |
true,
|
| 1278 |
true,
|
|
@@ -1296,9 +1296,8 @@
|
|
| 1296 |
false
|
| 1297 |
],
|
| 1298 |
"model.decoder.layers.17.encoder_attn.out_proj.lora_E": [
|
| 1299 |
-
true,
|
| 1300 |
false,
|
| 1301 |
-
|
| 1302 |
true,
|
| 1303 |
false,
|
| 1304 |
false,
|
|
@@ -1306,7 +1305,8 @@
|
|
| 1306 |
false,
|
| 1307 |
false,
|
| 1308 |
false,
|
| 1309 |
-
|
|
|
|
| 1310 |
false
|
| 1311 |
],
|
| 1312 |
"model.decoder.layers.17.encoder_attn.q_proj.lora_E": [
|
|
@@ -1329,7 +1329,7 @@
|
|
| 1329 |
true,
|
| 1330 |
true,
|
| 1331 |
true,
|
| 1332 |
-
|
| 1333 |
true,
|
| 1334 |
true,
|
| 1335 |
true,
|
|
@@ -1375,7 +1375,7 @@
|
|
| 1375 |
false,
|
| 1376 |
false,
|
| 1377 |
false,
|
| 1378 |
-
|
| 1379 |
false,
|
| 1380 |
false
|
| 1381 |
],
|
|
@@ -1447,7 +1447,7 @@
|
|
| 1447 |
false,
|
| 1448 |
false,
|
| 1449 |
false,
|
| 1450 |
-
|
| 1451 |
],
|
| 1452 |
"model.decoder.layers.18.encoder_attn.q_proj.lora_E": [
|
| 1453 |
false,
|
|
@@ -1461,7 +1461,7 @@
|
|
| 1461 |
false,
|
| 1462 |
false,
|
| 1463 |
false,
|
| 1464 |
-
|
| 1465 |
],
|
| 1466 |
"model.decoder.layers.18.encoder_attn.v_proj.lora_E": [
|
| 1467 |
true,
|
|
@@ -1503,7 +1503,7 @@
|
|
| 1503 |
true,
|
| 1504 |
true,
|
| 1505 |
false,
|
| 1506 |
-
|
| 1507 |
],
|
| 1508 |
"model.decoder.layers.18.self_attn.k_proj.lora_E": [
|
| 1509 |
false,
|
|
@@ -1517,11 +1517,11 @@
|
|
| 1517 |
false,
|
| 1518 |
false,
|
| 1519 |
false,
|
| 1520 |
-
|
| 1521 |
],
|
| 1522 |
"model.decoder.layers.18.self_attn.out_proj.lora_E": [
|
| 1523 |
true,
|
| 1524 |
-
|
| 1525 |
true,
|
| 1526 |
false,
|
| 1527 |
true,
|
|
@@ -1552,7 +1552,7 @@
|
|
| 1552 |
false,
|
| 1553 |
true,
|
| 1554 |
false,
|
| 1555 |
-
|
| 1556 |
true,
|
| 1557 |
false,
|
| 1558 |
false,
|
|
@@ -1578,12 +1578,12 @@
|
|
| 1578 |
"model.decoder.layers.19.encoder_attn.out_proj.lora_E": [
|
| 1579 |
false,
|
| 1580 |
false,
|
| 1581 |
-
true,
|
| 1582 |
false,
|
| 1583 |
false,
|
| 1584 |
false,
|
| 1585 |
false,
|
| 1586 |
-
|
|
|
|
| 1587 |
false,
|
| 1588 |
false,
|
| 1589 |
false,
|
|
@@ -1608,14 +1608,14 @@
|
|
| 1608 |
true,
|
| 1609 |
false,
|
| 1610 |
false,
|
| 1611 |
-
|
| 1612 |
false,
|
| 1613 |
true,
|
| 1614 |
false,
|
| 1615 |
false,
|
| 1616 |
false,
|
| 1617 |
false,
|
| 1618 |
-
|
| 1619 |
],
|
| 1620 |
"model.decoder.layers.19.fc1.lora_E": [
|
| 1621 |
true,
|
|
@@ -1636,7 +1636,7 @@
|
|
| 1636 |
false,
|
| 1637 |
true,
|
| 1638 |
false,
|
| 1639 |
-
|
| 1640 |
true,
|
| 1641 |
false,
|
| 1642 |
true,
|
|
@@ -1663,12 +1663,12 @@
|
|
| 1663 |
true,
|
| 1664 |
false,
|
| 1665 |
false,
|
| 1666 |
-
|
| 1667 |
false,
|
| 1668 |
true,
|
| 1669 |
false,
|
| 1670 |
false,
|
| 1671 |
-
|
| 1672 |
true,
|
| 1673 |
false,
|
| 1674 |
false
|
|
@@ -1699,7 +1699,7 @@
|
|
| 1699 |
false,
|
| 1700 |
true,
|
| 1701 |
false,
|
| 1702 |
-
|
| 1703 |
],
|
| 1704 |
"model.decoder.layers.2.encoder_attn.k_proj.lora_E": [
|
| 1705 |
false,
|
|
@@ -1726,7 +1726,7 @@
|
|
| 1726 |
false,
|
| 1727 |
false,
|
| 1728 |
false,
|
| 1729 |
-
|
| 1730 |
false
|
| 1731 |
],
|
| 1732 |
"model.decoder.layers.2.encoder_attn.q_proj.lora_E": [
|
|
@@ -1861,9 +1861,9 @@
|
|
| 1861 |
false,
|
| 1862 |
true,
|
| 1863 |
true,
|
|
|
|
| 1864 |
true,
|
| 1865 |
-
|
| 1866 |
-
true,
|
| 1867 |
true,
|
| 1868 |
true,
|
| 1869 |
true,
|
|
@@ -1922,11 +1922,11 @@
|
|
| 1922 |
true,
|
| 1923 |
true,
|
| 1924 |
true,
|
| 1925 |
-
|
| 1926 |
true
|
| 1927 |
],
|
| 1928 |
"model.decoder.layers.20.self_attn.k_proj.lora_E": [
|
| 1929 |
-
|
| 1930 |
true,
|
| 1931 |
true,
|
| 1932 |
true,
|
|
@@ -2005,8 +2005,8 @@
|
|
| 2005 |
false,
|
| 2006 |
true,
|
| 2007 |
true,
|
| 2008 |
-
|
| 2009 |
-
|
| 2010 |
true
|
| 2011 |
],
|
| 2012 |
"model.decoder.layers.21.encoder_attn.q_proj.lora_E": [
|
|
@@ -2056,7 +2056,7 @@
|
|
| 2056 |
false,
|
| 2057 |
true,
|
| 2058 |
true,
|
| 2059 |
-
|
| 2060 |
true,
|
| 2061 |
true,
|
| 2062 |
true,
|
|
@@ -2066,14 +2066,14 @@
|
|
| 2066 |
true
|
| 2067 |
],
|
| 2068 |
"model.decoder.layers.21.self_attn.k_proj.lora_E": [
|
| 2069 |
-
true,
|
| 2070 |
true,
|
| 2071 |
false,
|
| 2072 |
-
|
| 2073 |
true,
|
| 2074 |
true,
|
| 2075 |
false,
|
| 2076 |
false,
|
|
|
|
| 2077 |
true,
|
| 2078 |
true,
|
| 2079 |
true,
|
|
@@ -2100,7 +2100,7 @@
|
|
| 2100 |
false,
|
| 2101 |
false,
|
| 2102 |
true,
|
| 2103 |
-
|
| 2104 |
false,
|
| 2105 |
true,
|
| 2106 |
false,
|
|
@@ -2136,7 +2136,7 @@
|
|
| 2136 |
false
|
| 2137 |
],
|
| 2138 |
"model.decoder.layers.22.encoder_attn.out_proj.lora_E": [
|
| 2139 |
-
|
| 2140 |
true,
|
| 2141 |
false,
|
| 2142 |
true,
|
|
@@ -2220,13 +2220,13 @@
|
|
| 2220 |
false
|
| 2221 |
],
|
| 2222 |
"model.decoder.layers.22.self_attn.out_proj.lora_E": [
|
| 2223 |
-
false,
|
| 2224 |
-
false,
|
| 2225 |
false,
|
| 2226 |
true,
|
| 2227 |
false,
|
|
|
|
| 2228 |
false,
|
| 2229 |
true,
|
|
|
|
| 2230 |
true,
|
| 2231 |
true,
|
| 2232 |
false,
|
|
@@ -2243,7 +2243,7 @@
|
|
| 2243 |
false,
|
| 2244 |
false,
|
| 2245 |
false,
|
| 2246 |
-
|
| 2247 |
false,
|
| 2248 |
false
|
| 2249 |
],
|
|
@@ -2257,9 +2257,9 @@
|
|
| 2257 |
true,
|
| 2258 |
true,
|
| 2259 |
true,
|
| 2260 |
-
|
| 2261 |
-
|
| 2262 |
-
|
| 2263 |
],
|
| 2264 |
"model.decoder.layers.23.encoder_attn.k_proj.lora_E": [
|
| 2265 |
false,
|
|
@@ -2347,15 +2347,15 @@
|
|
| 2347 |
],
|
| 2348 |
"model.decoder.layers.23.self_attn.k_proj.lora_E": [
|
| 2349 |
false,
|
| 2350 |
-
true,
|
| 2351 |
false,
|
| 2352 |
false,
|
| 2353 |
true,
|
|
|
|
| 2354 |
false,
|
| 2355 |
false,
|
| 2356 |
true,
|
| 2357 |
false,
|
| 2358 |
-
|
| 2359 |
false,
|
| 2360 |
true
|
| 2361 |
],
|
|
@@ -2364,7 +2364,7 @@
|
|
| 2364 |
true,
|
| 2365 |
true,
|
| 2366 |
true,
|
| 2367 |
-
|
| 2368 |
false,
|
| 2369 |
false,
|
| 2370 |
true,
|
|
@@ -2395,9 +2395,9 @@
|
|
| 2395 |
true,
|
| 2396 |
false,
|
| 2397 |
true,
|
| 2398 |
-
false,
|
| 2399 |
true,
|
| 2400 |
-
|
|
|
|
| 2401 |
false,
|
| 2402 |
true
|
| 2403 |
],
|
|
@@ -2481,9 +2481,9 @@
|
|
| 2481 |
true,
|
| 2482 |
true,
|
| 2483 |
true,
|
| 2484 |
-
true,
|
| 2485 |
false,
|
| 2486 |
-
false
|
|
|
|
| 2487 |
],
|
| 2488 |
"model.decoder.layers.24.self_attn.k_proj.lora_E": [
|
| 2489 |
true,
|
|
@@ -2492,12 +2492,12 @@
|
|
| 2492 |
true,
|
| 2493 |
true,
|
| 2494 |
true,
|
|
|
|
| 2495 |
true,
|
| 2496 |
true,
|
| 2497 |
true,
|
| 2498 |
true,
|
| 2499 |
-
|
| 2500 |
-
true
|
| 2501 |
],
|
| 2502 |
"model.decoder.layers.24.self_attn.out_proj.lora_E": [
|
| 2503 |
true,
|
|
@@ -2558,10 +2558,10 @@
|
|
| 2558 |
"model.decoder.layers.25.encoder_attn.out_proj.lora_E": [
|
| 2559 |
true,
|
| 2560 |
true,
|
| 2561 |
-
|
| 2562 |
-
true,
|
| 2563 |
true,
|
| 2564 |
false,
|
|
|
|
| 2565 |
true,
|
| 2566 |
true,
|
| 2567 |
true,
|
|
@@ -2626,15 +2626,15 @@
|
|
| 2626 |
true
|
| 2627 |
],
|
| 2628 |
"model.decoder.layers.25.self_attn.k_proj.lora_E": [
|
| 2629 |
-
true,
|
| 2630 |
-
true,
|
| 2631 |
-
true,
|
| 2632 |
true,
|
| 2633 |
false,
|
| 2634 |
true,
|
| 2635 |
true,
|
|
|
|
| 2636 |
true,
|
| 2637 |
true,
|
|
|
|
|
|
|
| 2638 |
true,
|
| 2639 |
true,
|
| 2640 |
false
|
|
@@ -2649,7 +2649,7 @@
|
|
| 2649 |
false,
|
| 2650 |
false,
|
| 2651 |
true,
|
| 2652 |
-
|
| 2653 |
false,
|
| 2654 |
false
|
| 2655 |
],
|
|
@@ -2698,7 +2698,7 @@
|
|
| 2698 |
"model.decoder.layers.26.encoder_attn.out_proj.lora_E": [
|
| 2699 |
true,
|
| 2700 |
true,
|
| 2701 |
-
|
| 2702 |
true,
|
| 2703 |
true,
|
| 2704 |
true,
|
|
@@ -2706,7 +2706,7 @@
|
|
| 2706 |
true,
|
| 2707 |
true,
|
| 2708 |
true,
|
| 2709 |
-
|
| 2710 |
true
|
| 2711 |
],
|
| 2712 |
"model.decoder.layers.26.encoder_attn.q_proj.lora_E": [
|
|
@@ -2766,7 +2766,7 @@
|
|
| 2766 |
false
|
| 2767 |
],
|
| 2768 |
"model.decoder.layers.26.self_attn.k_proj.lora_E": [
|
| 2769 |
-
|
| 2770 |
true,
|
| 2771 |
false,
|
| 2772 |
true,
|
|
@@ -2774,18 +2774,18 @@
|
|
| 2774 |
false,
|
| 2775 |
false,
|
| 2776 |
false,
|
| 2777 |
-
|
| 2778 |
true,
|
| 2779 |
false,
|
| 2780 |
false
|
| 2781 |
],
|
| 2782 |
"model.decoder.layers.26.self_attn.out_proj.lora_E": [
|
| 2783 |
-
false,
|
| 2784 |
false,
|
| 2785 |
false,
|
| 2786 |
true,
|
| 2787 |
-
false,
|
| 2788 |
true,
|
|
|
|
|
|
|
| 2789 |
true,
|
| 2790 |
false,
|
| 2791 |
false,
|
|
@@ -2794,9 +2794,7 @@
|
|
| 2794 |
true
|
| 2795 |
],
|
| 2796 |
"model.decoder.layers.26.self_attn.q_proj.lora_E": [
|
| 2797 |
-
true,
|
| 2798 |
false,
|
| 2799 |
-
true,
|
| 2800 |
false,
|
| 2801 |
false,
|
| 2802 |
false,
|
|
@@ -2805,19 +2803,21 @@
|
|
| 2805 |
false,
|
| 2806 |
false,
|
| 2807 |
false,
|
| 2808 |
-
|
|
|
|
|
|
|
| 2809 |
],
|
| 2810 |
"model.decoder.layers.26.self_attn.v_proj.lora_E": [
|
| 2811 |
false,
|
| 2812 |
true,
|
| 2813 |
false,
|
| 2814 |
-
true,
|
| 2815 |
-
true,
|
| 2816 |
false,
|
|
|
|
| 2817 |
false,
|
| 2818 |
false,
|
| 2819 |
true,
|
| 2820 |
true,
|
|
|
|
| 2821 |
false,
|
| 2822 |
true
|
| 2823 |
],
|
|
@@ -2845,7 +2845,7 @@
|
|
| 2845 |
true,
|
| 2846 |
true,
|
| 2847 |
true,
|
| 2848 |
-
|
| 2849 |
false,
|
| 2850 |
true
|
| 2851 |
],
|
|
@@ -2907,7 +2907,7 @@
|
|
| 2907 |
],
|
| 2908 |
"model.decoder.layers.27.self_attn.k_proj.lora_E": [
|
| 2909 |
true,
|
| 2910 |
-
|
| 2911 |
true,
|
| 2912 |
false,
|
| 2913 |
false,
|
|
@@ -2948,7 +2948,7 @@
|
|
| 2948 |
false
|
| 2949 |
],
|
| 2950 |
"model.decoder.layers.27.self_attn.v_proj.lora_E": [
|
| 2951 |
-
|
| 2952 |
true,
|
| 2953 |
false,
|
| 2954 |
false,
|
|
@@ -2977,13 +2977,13 @@
|
|
| 2977 |
],
|
| 2978 |
"model.decoder.layers.28.encoder_attn.out_proj.lora_E": [
|
| 2979 |
true,
|
|
|
|
| 2980 |
true,
|
| 2981 |
true,
|
| 2982 |
true,
|
| 2983 |
true,
|
| 2984 |
-
false,
|
| 2985 |
-
true,
|
| 2986 |
true,
|
|
|
|
| 2987 |
true,
|
| 2988 |
true,
|
| 2989 |
true,
|
|
@@ -3041,12 +3041,12 @@
|
|
| 3041 |
true,
|
| 3042 |
true,
|
| 3043 |
true,
|
| 3044 |
-
|
| 3045 |
true,
|
| 3046 |
true
|
| 3047 |
],
|
| 3048 |
"model.decoder.layers.28.self_attn.k_proj.lora_E": [
|
| 3049 |
-
|
| 3050 |
true,
|
| 3051 |
true,
|
| 3052 |
false,
|
|
@@ -3068,7 +3068,7 @@
|
|
| 3068 |
true,
|
| 3069 |
true,
|
| 3070 |
false,
|
| 3071 |
-
|
| 3072 |
false,
|
| 3073 |
false,
|
| 3074 |
false
|
|
@@ -3097,7 +3097,7 @@
|
|
| 3097 |
false,
|
| 3098 |
false,
|
| 3099 |
false,
|
| 3100 |
-
|
| 3101 |
false,
|
| 3102 |
true
|
| 3103 |
],
|
|
@@ -3116,18 +3116,18 @@
|
|
| 3116 |
false
|
| 3117 |
],
|
| 3118 |
"model.decoder.layers.29.encoder_attn.out_proj.lora_E": [
|
| 3119 |
-
true,
|
| 3120 |
-
true,
|
| 3121 |
true,
|
| 3122 |
false,
|
| 3123 |
false,
|
| 3124 |
false,
|
| 3125 |
-
true,
|
| 3126 |
false,
|
| 3127 |
false,
|
| 3128 |
false,
|
| 3129 |
false,
|
| 3130 |
-
|
|
|
|
|
|
|
|
|
|
| 3131 |
],
|
| 3132 |
"model.decoder.layers.29.encoder_attn.q_proj.lora_E": [
|
| 3133 |
false,
|
|
@@ -3191,7 +3191,7 @@
|
|
| 3191 |
false,
|
| 3192 |
false,
|
| 3193 |
false,
|
| 3194 |
-
|
| 3195 |
true,
|
| 3196 |
false,
|
| 3197 |
false,
|
|
@@ -3210,7 +3210,7 @@
|
|
| 3210 |
false,
|
| 3211 |
false,
|
| 3212 |
false,
|
| 3213 |
-
|
| 3214 |
false
|
| 3215 |
],
|
| 3216 |
"model.decoder.layers.29.self_attn.q_proj.lora_E": [
|
|
@@ -3259,13 +3259,13 @@
|
|
| 3259 |
true,
|
| 3260 |
true,
|
| 3261 |
false,
|
| 3262 |
-
true,
|
| 3263 |
-
false,
|
| 3264 |
false,
|
| 3265 |
true,
|
|
|
|
| 3266 |
true,
|
| 3267 |
true,
|
| 3268 |
false,
|
|
|
|
| 3269 |
true,
|
| 3270 |
false
|
| 3271 |
],
|
|
@@ -3368,7 +3368,7 @@
|
|
| 3368 |
false
|
| 3369 |
],
|
| 3370 |
"model.decoder.layers.3.self_attn.v_proj.lora_E": [
|
| 3371 |
-
|
| 3372 |
false,
|
| 3373 |
false,
|
| 3374 |
false,
|
|
@@ -3378,7 +3378,7 @@
|
|
| 3378 |
false,
|
| 3379 |
false,
|
| 3380 |
true,
|
| 3381 |
-
|
| 3382 |
false
|
| 3383 |
],
|
| 3384 |
"model.decoder.layers.30.encoder_attn.k_proj.lora_E": [
|
|
@@ -3396,17 +3396,17 @@
|
|
| 3396 |
false
|
| 3397 |
],
|
| 3398 |
"model.decoder.layers.30.encoder_attn.out_proj.lora_E": [
|
| 3399 |
-
true,
|
| 3400 |
true,
|
| 3401 |
false,
|
| 3402 |
false,
|
| 3403 |
false,
|
| 3404 |
false,
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3405 |
true,
|
| 3406 |
-
|
| 3407 |
-
true,
|
| 3408 |
-
true,
|
| 3409 |
-
true,
|
| 3410 |
false
|
| 3411 |
],
|
| 3412 |
"model.decoder.layers.30.encoder_attn.q_proj.lora_E": [
|
|
@@ -3475,7 +3475,7 @@
|
|
| 3475 |
false,
|
| 3476 |
true,
|
| 3477 |
false,
|
| 3478 |
-
|
| 3479 |
true,
|
| 3480 |
true
|
| 3481 |
],
|
|
@@ -3517,7 +3517,7 @@
|
|
| 3517 |
false,
|
| 3518 |
false,
|
| 3519 |
false,
|
| 3520 |
-
|
| 3521 |
false,
|
| 3522 |
true
|
| 3523 |
],
|
|
@@ -3584,7 +3584,7 @@
|
|
| 3584 |
false,
|
| 3585 |
false,
|
| 3586 |
false,
|
| 3587 |
-
|
| 3588 |
false,
|
| 3589 |
true,
|
| 3590 |
true,
|
|
@@ -3686,7 +3686,7 @@
|
|
| 3686 |
false,
|
| 3687 |
false,
|
| 3688 |
false,
|
| 3689 |
-
|
| 3690 |
false
|
| 3691 |
],
|
| 3692 |
"model.decoder.layers.4.encoder_attn.q_proj.lora_E": [
|
|
@@ -3705,7 +3705,7 @@
|
|
| 3705 |
],
|
| 3706 |
"model.decoder.layers.4.encoder_attn.v_proj.lora_E": [
|
| 3707 |
true,
|
| 3708 |
-
|
| 3709 |
false,
|
| 3710 |
false,
|
| 3711 |
false,
|
|
@@ -3714,8 +3714,8 @@
|
|
| 3714 |
true,
|
| 3715 |
false,
|
| 3716 |
true,
|
| 3717 |
-
|
| 3718 |
-
|
| 3719 |
],
|
| 3720 |
"model.decoder.layers.4.fc1.lora_E": [
|
| 3721 |
true,
|
|
@@ -3853,7 +3853,7 @@
|
|
| 3853 |
false,
|
| 3854 |
true,
|
| 3855 |
true,
|
| 3856 |
-
|
| 3857 |
false,
|
| 3858 |
true
|
| 3859 |
],
|
|
@@ -4048,7 +4048,7 @@
|
|
| 4048 |
false,
|
| 4049 |
false,
|
| 4050 |
false,
|
| 4051 |
-
|
| 4052 |
false,
|
| 4053 |
false,
|
| 4054 |
false
|
|
@@ -4069,7 +4069,7 @@
|
|
| 4069 |
],
|
| 4070 |
"model.decoder.layers.6.self_attn.v_proj.lora_E": [
|
| 4071 |
false,
|
| 4072 |
-
|
| 4073 |
false,
|
| 4074 |
false,
|
| 4075 |
false,
|
|
@@ -4079,7 +4079,7 @@
|
|
| 4079 |
false,
|
| 4080 |
true,
|
| 4081 |
false,
|
| 4082 |
-
|
| 4083 |
],
|
| 4084 |
"model.decoder.layers.7.encoder_attn.k_proj.lora_E": [
|
| 4085 |
false,
|
|
@@ -4149,7 +4149,7 @@
|
|
| 4149 |
true,
|
| 4150 |
true,
|
| 4151 |
true,
|
| 4152 |
-
|
| 4153 |
],
|
| 4154 |
"model.decoder.layers.7.fc2.lora_E": [
|
| 4155 |
true,
|
|
@@ -4428,7 +4428,7 @@
|
|
| 4428 |
true,
|
| 4429 |
true,
|
| 4430 |
true,
|
| 4431 |
-
|
| 4432 |
true
|
| 4433 |
],
|
| 4434 |
"model.decoder.layers.9.fc2.lora_E": [
|
|
@@ -4443,7 +4443,7 @@
|
|
| 4443 |
true,
|
| 4444 |
true,
|
| 4445 |
true,
|
| 4446 |
-
|
| 4447 |
],
|
| 4448 |
"model.decoder.layers.9.self_attn.k_proj.lora_E": [
|
| 4449 |
false,
|
|
@@ -4496,9 +4496,9 @@
|
|
| 4496 |
true,
|
| 4497 |
false,
|
| 4498 |
false,
|
| 4499 |
-
true,
|
| 4500 |
false,
|
| 4501 |
false,
|
|
|
|
| 4502 |
false
|
| 4503 |
],
|
| 4504 |
"model.encoder.layers.0.fc1.lora_E": [
|
|
@@ -4550,8 +4550,8 @@
|
|
| 4550 |
true,
|
| 4551 |
false,
|
| 4552 |
false,
|
| 4553 |
-
|
| 4554 |
-
|
| 4555 |
false,
|
| 4556 |
false,
|
| 4557 |
false,
|
|
@@ -4778,7 +4778,7 @@
|
|
| 4778 |
false,
|
| 4779 |
true,
|
| 4780 |
false,
|
| 4781 |
-
|
| 4782 |
false
|
| 4783 |
],
|
| 4784 |
"model.encoder.layers.11.self_attn.k_proj.lora_E": [
|
|
@@ -4832,7 +4832,7 @@
|
|
| 4832 |
false,
|
| 4833 |
false,
|
| 4834 |
false,
|
| 4835 |
-
|
| 4836 |
false,
|
| 4837 |
false,
|
| 4838 |
false
|
|
@@ -4885,7 +4885,7 @@
|
|
| 4885 |
false,
|
| 4886 |
false,
|
| 4887 |
false,
|
| 4888 |
-
|
| 4889 |
false,
|
| 4890 |
false,
|
| 4891 |
false,
|
|
@@ -4926,7 +4926,7 @@
|
|
| 4926 |
false,
|
| 4927 |
false,
|
| 4928 |
false,
|
| 4929 |
-
|
| 4930 |
true,
|
| 4931 |
false,
|
| 4932 |
false,
|
|
@@ -4943,7 +4943,7 @@
|
|
| 4943 |
true,
|
| 4944 |
false,
|
| 4945 |
false,
|
| 4946 |
-
|
| 4947 |
false,
|
| 4948 |
false,
|
| 4949 |
true,
|
|
@@ -4967,7 +4967,7 @@
|
|
| 4967 |
false,
|
| 4968 |
false,
|
| 4969 |
false,
|
| 4970 |
-
|
| 4971 |
false,
|
| 4972 |
false,
|
| 4973 |
false,
|
|
@@ -5003,7 +5003,7 @@
|
|
| 5003 |
false,
|
| 5004 |
false,
|
| 5005 |
false,
|
| 5006 |
-
|
| 5007 |
],
|
| 5008 |
"model.encoder.layers.14.fc1.lora_E": [
|
| 5009 |
true,
|
|
@@ -5016,7 +5016,7 @@
|
|
| 5016 |
true,
|
| 5017 |
false,
|
| 5018 |
false,
|
| 5019 |
-
|
| 5020 |
true
|
| 5021 |
],
|
| 5022 |
"model.encoder.layers.14.fc2.lora_E": [
|
|
@@ -5056,7 +5056,7 @@
|
|
| 5056 |
false,
|
| 5057 |
false,
|
| 5058 |
true,
|
| 5059 |
-
|
| 5060 |
false,
|
| 5061 |
true,
|
| 5062 |
false
|
|
@@ -5136,11 +5136,11 @@
|
|
| 5136 |
true,
|
| 5137 |
false,
|
| 5138 |
false,
|
| 5139 |
-
|
| 5140 |
true,
|
| 5141 |
false,
|
| 5142 |
false,
|
| 5143 |
-
|
| 5144 |
false,
|
| 5145 |
false,
|
| 5146 |
false
|
|
@@ -5168,7 +5168,7 @@
|
|
| 5168 |
false,
|
| 5169 |
false,
|
| 5170 |
true,
|
| 5171 |
-
|
| 5172 |
false,
|
| 5173 |
false,
|
| 5174 |
true
|
|
@@ -5223,10 +5223,10 @@
|
|
| 5223 |
true,
|
| 5224 |
false,
|
| 5225 |
true,
|
| 5226 |
-
true,
|
| 5227 |
false,
|
| 5228 |
false,
|
| 5229 |
false,
|
|
|
|
| 5230 |
true
|
| 5231 |
],
|
| 5232 |
"model.encoder.layers.16.self_attn.q_proj.lora_E": [
|
|
@@ -5307,9 +5307,9 @@
|
|
| 5307 |
true,
|
| 5308 |
true,
|
| 5309 |
false,
|
|
|
|
| 5310 |
true,
|
| 5311 |
-
|
| 5312 |
-
true,
|
| 5313 |
true,
|
| 5314 |
true
|
| 5315 |
],
|
|
@@ -5386,12 +5386,12 @@
|
|
| 5386 |
"model.encoder.layers.18.self_attn.out_proj.lora_E": [
|
| 5387 |
false,
|
| 5388 |
false,
|
| 5389 |
-
true,
|
| 5390 |
false,
|
| 5391 |
false,
|
| 5392 |
false,
|
| 5393 |
-
|
| 5394 |
-
|
|
|
|
| 5395 |
true,
|
| 5396 |
false,
|
| 5397 |
false,
|
|
@@ -5474,7 +5474,7 @@
|
|
| 5474 |
false,
|
| 5475 |
false,
|
| 5476 |
false,
|
| 5477 |
-
|
| 5478 |
false,
|
| 5479 |
false,
|
| 5480 |
false,
|
|
@@ -5806,12 +5806,12 @@
|
|
| 5806 |
"model.encoder.layers.22.self_attn.out_proj.lora_E": [
|
| 5807 |
false,
|
| 5808 |
false,
|
| 5809 |
-
|
| 5810 |
false,
|
| 5811 |
false,
|
| 5812 |
true,
|
| 5813 |
-
|
| 5814 |
-
|
| 5815 |
true,
|
| 5816 |
false,
|
| 5817 |
false,
|
|
@@ -5892,13 +5892,13 @@
|
|
| 5892 |
true,
|
| 5893 |
true,
|
| 5894 |
true,
|
| 5895 |
-
|
| 5896 |
-
|
| 5897 |
true,
|
| 5898 |
true,
|
| 5899 |
false,
|
| 5900 |
false,
|
| 5901 |
-
|
| 5902 |
true
|
| 5903 |
],
|
| 5904 |
"model.encoder.layers.23.self_attn.q_proj.lora_E": [
|
|
@@ -5916,10 +5916,10 @@
|
|
| 5916 |
false
|
| 5917 |
],
|
| 5918 |
"model.encoder.layers.23.self_attn.v_proj.lora_E": [
|
| 5919 |
-
true,
|
| 5920 |
true,
|
| 5921 |
false,
|
| 5922 |
false,
|
|
|
|
| 5923 |
true,
|
| 5924 |
true,
|
| 5925 |
true,
|
|
@@ -6011,7 +6011,7 @@
|
|
| 6011 |
true,
|
| 6012 |
true,
|
| 6013 |
true,
|
| 6014 |
-
|
| 6015 |
],
|
| 6016 |
"model.encoder.layers.25.fc1.lora_E": [
|
| 6017 |
true,
|
|
@@ -6092,7 +6092,7 @@
|
|
| 6092 |
true,
|
| 6093 |
true,
|
| 6094 |
true,
|
| 6095 |
-
|
| 6096 |
true,
|
| 6097 |
true,
|
| 6098 |
true
|
|
@@ -6227,11 +6227,11 @@
|
|
| 6227 |
true,
|
| 6228 |
true,
|
| 6229 |
false,
|
| 6230 |
-
false,
|
| 6231 |
-
false,
|
| 6232 |
true,
|
|
|
|
| 6233 |
true,
|
| 6234 |
true,
|
|
|
|
| 6235 |
true,
|
| 6236 |
false,
|
| 6237 |
false,
|
|
@@ -6309,7 +6309,7 @@
|
|
| 6309 |
],
|
| 6310 |
"model.encoder.layers.28.self_attn.out_proj.lora_E": [
|
| 6311 |
false,
|
| 6312 |
-
|
| 6313 |
false,
|
| 6314 |
true,
|
| 6315 |
false,
|
|
@@ -6319,7 +6319,7 @@
|
|
| 6319 |
true,
|
| 6320 |
true,
|
| 6321 |
true,
|
| 6322 |
-
|
| 6323 |
],
|
| 6324 |
"model.encoder.layers.28.self_attn.q_proj.lora_E": [
|
| 6325 |
false,
|
|
@@ -6400,7 +6400,7 @@
|
|
| 6400 |
true,
|
| 6401 |
true,
|
| 6402 |
true,
|
| 6403 |
-
|
| 6404 |
true,
|
| 6405 |
true,
|
| 6406 |
true
|
|
@@ -6425,7 +6425,7 @@
|
|
| 6425 |
true,
|
| 6426 |
true,
|
| 6427 |
true,
|
| 6428 |
-
|
| 6429 |
true,
|
| 6430 |
true,
|
| 6431 |
true,
|
|
@@ -6564,12 +6564,12 @@
|
|
| 6564 |
true,
|
| 6565 |
true,
|
| 6566 |
true,
|
| 6567 |
-
false,
|
| 6568 |
true,
|
| 6569 |
true,
|
| 6570 |
true,
|
| 6571 |
true,
|
| 6572 |
true,
|
|
|
|
| 6573 |
true,
|
| 6574 |
false
|
| 6575 |
],
|
|
@@ -6649,7 +6649,7 @@
|
|
| 6649 |
false,
|
| 6650 |
false,
|
| 6651 |
true,
|
| 6652 |
-
|
| 6653 |
true,
|
| 6654 |
true,
|
| 6655 |
true,
|
|
@@ -6677,7 +6677,7 @@
|
|
| 6677 |
true,
|
| 6678 |
true,
|
| 6679 |
false,
|
| 6680 |
-
|
| 6681 |
false,
|
| 6682 |
true,
|
| 6683 |
true,
|
|
@@ -6794,7 +6794,7 @@
|
|
| 6794 |
false,
|
| 6795 |
false,
|
| 6796 |
false,
|
| 6797 |
-
|
| 6798 |
false
|
| 6799 |
],
|
| 6800 |
"model.encoder.layers.5.self_attn.k_proj.lora_E": [
|
|
@@ -6871,7 +6871,7 @@
|
|
| 6871 |
false,
|
| 6872 |
false,
|
| 6873 |
false,
|
| 6874 |
-
|
| 6875 |
false,
|
| 6876 |
false,
|
| 6877 |
false,
|
|
@@ -7040,7 +7040,7 @@
|
|
| 7040 |
false,
|
| 7041 |
false,
|
| 7042 |
false,
|
| 7043 |
-
|
| 7044 |
false,
|
| 7045 |
false,
|
| 7046 |
false,
|
|
@@ -7129,8 +7129,8 @@
|
|
| 7129 |
false,
|
| 7130 |
false,
|
| 7131 |
false,
|
| 7132 |
-
|
| 7133 |
-
|
| 7134 |
false
|
| 7135 |
],
|
| 7136 |
"model.encoder.layers.9.self_attn.k_proj.lora_E": [
|
|
|
|
| 38 |
"model.decoder.layers.0.encoder_attn.out_proj.lora_E": [
|
| 39 |
true,
|
| 40 |
false,
|
| 41 |
+
false,
|
| 42 |
false,
|
| 43 |
false,
|
| 44 |
true,
|
|
|
|
| 73 |
false,
|
| 74 |
false,
|
| 75 |
false,
|
| 76 |
+
true,
|
| 77 |
true,
|
| 78 |
true
|
| 79 |
],
|
|
|
|
| 102 |
true,
|
| 103 |
true,
|
| 104 |
true,
|
| 105 |
+
true,
|
| 106 |
true
|
| 107 |
],
|
| 108 |
"model.decoder.layers.0.self_attn.k_proj.lora_E": [
|
|
|
|
| 215 |
false,
|
| 216 |
false,
|
| 217 |
false,
|
| 218 |
+
false
|
| 219 |
],
|
| 220 |
"model.decoder.layers.1.fc1.lora_E": [
|
| 221 |
true,
|
|
|
|
| 365 |
true,
|
| 366 |
true,
|
| 367 |
true,
|
| 368 |
+
false,
|
| 369 |
true,
|
| 370 |
true,
|
| 371 |
true,
|
|
|
|
| 504 |
true,
|
| 505 |
true,
|
| 506 |
true,
|
|
|
|
| 507 |
true,
|
| 508 |
true,
|
| 509 |
true,
|
| 510 |
true,
|
| 511 |
+
true,
|
| 512 |
+
false
|
| 513 |
],
|
| 514 |
"model.decoder.layers.11.fc2.lora_E": [
|
| 515 |
true,
|
|
|
|
| 550 |
false,
|
| 551 |
false,
|
| 552 |
false,
|
| 553 |
+
true,
|
| 554 |
true
|
| 555 |
],
|
| 556 |
"model.decoder.layers.11.self_attn.q_proj.lora_E": [
|
|
|
|
| 574 |
false,
|
| 575 |
false,
|
| 576 |
false,
|
| 577 |
+
false,
|
| 578 |
false,
|
| 579 |
false,
|
| 580 |
false,
|
|
|
|
| 680 |
false
|
| 681 |
],
|
| 682 |
"model.decoder.layers.12.self_attn.out_proj.lora_E": [
|
|
|
|
| 683 |
false,
|
| 684 |
false,
|
|
|
|
| 685 |
false,
|
| 686 |
false,
|
| 687 |
false,
|
| 688 |
false,
|
|
|
|
| 689 |
false,
|
| 690 |
false,
|
| 691 |
+
true,
|
| 692 |
+
false,
|
| 693 |
+
true,
|
| 694 |
false
|
| 695 |
],
|
| 696 |
"model.decoder.layers.12.self_attn.q_proj.lora_E": [
|
|
|
|
| 737 |
],
|
| 738 |
"model.decoder.layers.13.encoder_attn.out_proj.lora_E": [
|
| 739 |
false,
|
| 740 |
+
false,
|
| 741 |
false,
|
| 742 |
true,
|
| 743 |
false,
|
| 744 |
false,
|
| 745 |
false,
|
| 746 |
false,
|
|
|
|
| 747 |
false,
|
| 748 |
+
false,
|
| 749 |
+
false,
|
| 750 |
+
false
|
| 751 |
],
|
| 752 |
"model.decoder.layers.13.encoder_attn.q_proj.lora_E": [
|
| 753 |
false,
|
|
|
|
| 765 |
],
|
| 766 |
"model.decoder.layers.13.encoder_attn.v_proj.lora_E": [
|
| 767 |
false,
|
| 768 |
+
false,
|
| 769 |
false,
|
| 770 |
false,
|
| 771 |
true,
|
| 772 |
true,
|
| 773 |
true,
|
| 774 |
+
false,
|
| 775 |
+
false,
|
| 776 |
true,
|
| 777 |
true,
|
| 778 |
true
|
|
|
|
| 823 |
false,
|
| 824 |
false,
|
| 825 |
true,
|
| 826 |
+
false,
|
| 827 |
true,
|
| 828 |
false,
|
| 829 |
true,
|
|
|
|
| 859 |
false,
|
| 860 |
false,
|
| 861 |
false,
|
| 862 |
+
false
|
| 863 |
],
|
| 864 |
"model.decoder.layers.14.encoder_attn.k_proj.lora_E": [
|
| 865 |
false,
|
|
|
|
| 927 |
true,
|
| 928 |
true,
|
| 929 |
true,
|
| 930 |
+
false,
|
| 931 |
true,
|
| 932 |
true
|
| 933 |
],
|
|
|
|
| 962 |
"model.decoder.layers.14.self_attn.out_proj.lora_E": [
|
| 963 |
false,
|
| 964 |
true,
|
| 965 |
+
true,
|
| 966 |
true,
|
| 967 |
false,
|
| 968 |
false,
|
|
|
|
| 1077 |
true,
|
| 1078 |
true,
|
| 1079 |
true,
|
| 1080 |
+
false,
|
| 1081 |
true,
|
| 1082 |
true,
|
| 1083 |
true,
|
|
|
|
| 1108 |
true,
|
| 1109 |
true,
|
| 1110 |
false,
|
| 1111 |
+
true,
|
| 1112 |
false,
|
| 1113 |
false,
|
| 1114 |
true
|
|
|
|
| 1128 |
false
|
| 1129 |
],
|
| 1130 |
"model.decoder.layers.15.self_attn.v_proj.lora_E": [
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1131 |
true,
|
| 1132 |
false,
|
| 1133 |
+
true,
|
| 1134 |
false,
|
| 1135 |
+
true,
|
| 1136 |
false,
|
| 1137 |
false,
|
| 1138 |
true,
|
| 1139 |
true,
|
| 1140 |
+
true,
|
| 1141 |
+
true,
|
| 1142 |
false
|
| 1143 |
],
|
| 1144 |
"model.decoder.layers.16.encoder_attn.k_proj.lora_E": [
|
|
|
|
| 1181 |
false,
|
| 1182 |
false,
|
| 1183 |
false,
|
| 1184 |
+
false
|
| 1185 |
],
|
| 1186 |
"model.decoder.layers.16.encoder_attn.v_proj.lora_E": [
|
| 1187 |
true,
|
|
|
|
| 1221 |
true,
|
| 1222 |
true,
|
| 1223 |
true,
|
| 1224 |
+
true,
|
| 1225 |
true,
|
| 1226 |
true
|
| 1227 |
],
|
|
|
|
| 1272 |
true,
|
| 1273 |
true,
|
| 1274 |
true,
|
| 1275 |
+
true,
|
| 1276 |
true,
|
| 1277 |
true,
|
| 1278 |
true,
|
|
|
|
| 1296 |
false
|
| 1297 |
],
|
| 1298 |
"model.decoder.layers.17.encoder_attn.out_proj.lora_E": [
|
|
|
|
| 1299 |
false,
|
| 1300 |
+
false,
|
| 1301 |
true,
|
| 1302 |
false,
|
| 1303 |
false,
|
|
|
|
| 1305 |
false,
|
| 1306 |
false,
|
| 1307 |
false,
|
| 1308 |
+
false,
|
| 1309 |
+
false,
|
| 1310 |
false
|
| 1311 |
],
|
| 1312 |
"model.decoder.layers.17.encoder_attn.q_proj.lora_E": [
|
|
|
|
| 1329 |
true,
|
| 1330 |
true,
|
| 1331 |
true,
|
| 1332 |
+
true,
|
| 1333 |
true,
|
| 1334 |
true,
|
| 1335 |
true,
|
|
|
|
| 1375 |
false,
|
| 1376 |
false,
|
| 1377 |
false,
|
| 1378 |
+
false,
|
| 1379 |
false,
|
| 1380 |
false
|
| 1381 |
],
|
|
|
|
| 1447 |
false,
|
| 1448 |
false,
|
| 1449 |
false,
|
| 1450 |
+
false
|
| 1451 |
],
|
| 1452 |
"model.decoder.layers.18.encoder_attn.q_proj.lora_E": [
|
| 1453 |
false,
|
|
|
|
| 1461 |
false,
|
| 1462 |
false,
|
| 1463 |
false,
|
| 1464 |
+
false
|
| 1465 |
],
|
| 1466 |
"model.decoder.layers.18.encoder_attn.v_proj.lora_E": [
|
| 1467 |
true,
|
|
|
|
| 1503 |
true,
|
| 1504 |
true,
|
| 1505 |
false,
|
| 1506 |
+
true
|
| 1507 |
],
|
| 1508 |
"model.decoder.layers.18.self_attn.k_proj.lora_E": [
|
| 1509 |
false,
|
|
|
|
| 1517 |
false,
|
| 1518 |
false,
|
| 1519 |
false,
|
| 1520 |
+
false
|
| 1521 |
],
|
| 1522 |
"model.decoder.layers.18.self_attn.out_proj.lora_E": [
|
| 1523 |
true,
|
| 1524 |
+
false,
|
| 1525 |
true,
|
| 1526 |
false,
|
| 1527 |
true,
|
|
|
|
| 1552 |
false,
|
| 1553 |
true,
|
| 1554 |
false,
|
| 1555 |
+
true,
|
| 1556 |
true,
|
| 1557 |
false,
|
| 1558 |
false,
|
|
|
|
| 1578 |
"model.decoder.layers.19.encoder_attn.out_proj.lora_E": [
|
| 1579 |
false,
|
| 1580 |
false,
|
|
|
|
| 1581 |
false,
|
| 1582 |
false,
|
| 1583 |
false,
|
| 1584 |
false,
|
| 1585 |
+
false,
|
| 1586 |
+
false,
|
| 1587 |
false,
|
| 1588 |
false,
|
| 1589 |
false,
|
|
|
|
| 1608 |
true,
|
| 1609 |
false,
|
| 1610 |
false,
|
| 1611 |
+
false,
|
| 1612 |
false,
|
| 1613 |
true,
|
| 1614 |
false,
|
| 1615 |
false,
|
| 1616 |
false,
|
| 1617 |
false,
|
| 1618 |
+
false
|
| 1619 |
],
|
| 1620 |
"model.decoder.layers.19.fc1.lora_E": [
|
| 1621 |
true,
|
|
|
|
| 1636 |
false,
|
| 1637 |
true,
|
| 1638 |
false,
|
| 1639 |
+
false,
|
| 1640 |
true,
|
| 1641 |
false,
|
| 1642 |
true,
|
|
|
|
| 1663 |
true,
|
| 1664 |
false,
|
| 1665 |
false,
|
| 1666 |
+
true,
|
| 1667 |
false,
|
| 1668 |
true,
|
| 1669 |
false,
|
| 1670 |
false,
|
| 1671 |
+
false,
|
| 1672 |
true,
|
| 1673 |
false,
|
| 1674 |
false
|
|
|
|
| 1699 |
false,
|
| 1700 |
true,
|
| 1701 |
false,
|
| 1702 |
+
false
|
| 1703 |
],
|
| 1704 |
"model.decoder.layers.2.encoder_attn.k_proj.lora_E": [
|
| 1705 |
false,
|
|
|
|
| 1726 |
false,
|
| 1727 |
false,
|
| 1728 |
false,
|
| 1729 |
+
false,
|
| 1730 |
false
|
| 1731 |
],
|
| 1732 |
"model.decoder.layers.2.encoder_attn.q_proj.lora_E": [
|
|
|
|
| 1861 |
false,
|
| 1862 |
true,
|
| 1863 |
true,
|
| 1864 |
+
false,
|
| 1865 |
true,
|
| 1866 |
+
false,
|
|
|
|
| 1867 |
true,
|
| 1868 |
true,
|
| 1869 |
true,
|
|
|
|
| 1922 |
true,
|
| 1923 |
true,
|
| 1924 |
true,
|
| 1925 |
+
false,
|
| 1926 |
true
|
| 1927 |
],
|
| 1928 |
"model.decoder.layers.20.self_attn.k_proj.lora_E": [
|
| 1929 |
+
true,
|
| 1930 |
true,
|
| 1931 |
true,
|
| 1932 |
true,
|
|
|
|
| 2005 |
false,
|
| 2006 |
true,
|
| 2007 |
true,
|
| 2008 |
+
false,
|
| 2009 |
+
false,
|
| 2010 |
true
|
| 2011 |
],
|
| 2012 |
"model.decoder.layers.21.encoder_attn.q_proj.lora_E": [
|
|
|
|
| 2056 |
false,
|
| 2057 |
true,
|
| 2058 |
true,
|
| 2059 |
+
false,
|
| 2060 |
true,
|
| 2061 |
true,
|
| 2062 |
true,
|
|
|
|
| 2066 |
true
|
| 2067 |
],
|
| 2068 |
"model.decoder.layers.21.self_attn.k_proj.lora_E": [
|
|
|
|
| 2069 |
true,
|
| 2070 |
false,
|
| 2071 |
+
false,
|
| 2072 |
true,
|
| 2073 |
true,
|
| 2074 |
false,
|
| 2075 |
false,
|
| 2076 |
+
false,
|
| 2077 |
true,
|
| 2078 |
true,
|
| 2079 |
true,
|
|
|
|
| 2100 |
false,
|
| 2101 |
false,
|
| 2102 |
true,
|
| 2103 |
+
false,
|
| 2104 |
false,
|
| 2105 |
true,
|
| 2106 |
false,
|
|
|
|
| 2136 |
false
|
| 2137 |
],
|
| 2138 |
"model.decoder.layers.22.encoder_attn.out_proj.lora_E": [
|
| 2139 |
+
false,
|
| 2140 |
true,
|
| 2141 |
false,
|
| 2142 |
true,
|
|
|
|
| 2220 |
false
|
| 2221 |
],
|
| 2222 |
"model.decoder.layers.22.self_attn.out_proj.lora_E": [
|
|
|
|
|
|
|
| 2223 |
false,
|
| 2224 |
true,
|
| 2225 |
false,
|
| 2226 |
+
true,
|
| 2227 |
false,
|
| 2228 |
true,
|
| 2229 |
+
false,
|
| 2230 |
true,
|
| 2231 |
true,
|
| 2232 |
false,
|
|
|
|
| 2243 |
false,
|
| 2244 |
false,
|
| 2245 |
false,
|
| 2246 |
+
false,
|
| 2247 |
false,
|
| 2248 |
false
|
| 2249 |
],
|
|
|
|
| 2257 |
true,
|
| 2258 |
true,
|
| 2259 |
true,
|
| 2260 |
+
false,
|
| 2261 |
+
false,
|
| 2262 |
+
true
|
| 2263 |
],
|
| 2264 |
"model.decoder.layers.23.encoder_attn.k_proj.lora_E": [
|
| 2265 |
false,
|
|
|
|
| 2347 |
],
|
| 2348 |
"model.decoder.layers.23.self_attn.k_proj.lora_E": [
|
| 2349 |
false,
|
|
|
|
| 2350 |
false,
|
| 2351 |
false,
|
| 2352 |
true,
|
| 2353 |
+
true,
|
| 2354 |
false,
|
| 2355 |
false,
|
| 2356 |
true,
|
| 2357 |
false,
|
| 2358 |
+
true,
|
| 2359 |
false,
|
| 2360 |
true
|
| 2361 |
],
|
|
|
|
| 2364 |
true,
|
| 2365 |
true,
|
| 2366 |
true,
|
| 2367 |
+
true,
|
| 2368 |
false,
|
| 2369 |
false,
|
| 2370 |
true,
|
|
|
|
| 2395 |
true,
|
| 2396 |
false,
|
| 2397 |
true,
|
|
|
|
| 2398 |
true,
|
| 2399 |
+
true,
|
| 2400 |
+
true,
|
| 2401 |
false,
|
| 2402 |
true
|
| 2403 |
],
|
|
|
|
| 2481 |
true,
|
| 2482 |
true,
|
| 2483 |
true,
|
|
|
|
| 2484 |
false,
|
| 2485 |
+
false,
|
| 2486 |
+
true
|
| 2487 |
],
|
| 2488 |
"model.decoder.layers.24.self_attn.k_proj.lora_E": [
|
| 2489 |
true,
|
|
|
|
| 2492 |
true,
|
| 2493 |
true,
|
| 2494 |
true,
|
| 2495 |
+
false,
|
| 2496 |
true,
|
| 2497 |
true,
|
| 2498 |
true,
|
| 2499 |
true,
|
| 2500 |
+
false
|
|
|
|
| 2501 |
],
|
| 2502 |
"model.decoder.layers.24.self_attn.out_proj.lora_E": [
|
| 2503 |
true,
|
|
|
|
| 2558 |
"model.decoder.layers.25.encoder_attn.out_proj.lora_E": [
|
| 2559 |
true,
|
| 2560 |
true,
|
| 2561 |
+
false,
|
|
|
|
| 2562 |
true,
|
| 2563 |
false,
|
| 2564 |
+
false,
|
| 2565 |
true,
|
| 2566 |
true,
|
| 2567 |
true,
|
|
|
|
| 2626 |
true
|
| 2627 |
],
|
| 2628 |
"model.decoder.layers.25.self_attn.k_proj.lora_E": [
|
|
|
|
|
|
|
|
|
|
| 2629 |
true,
|
| 2630 |
false,
|
| 2631 |
true,
|
| 2632 |
true,
|
| 2633 |
+
false,
|
| 2634 |
true,
|
| 2635 |
true,
|
| 2636 |
+
false,
|
| 2637 |
+
false,
|
| 2638 |
true,
|
| 2639 |
true,
|
| 2640 |
false
|
|
|
|
| 2649 |
false,
|
| 2650 |
false,
|
| 2651 |
true,
|
| 2652 |
+
false,
|
| 2653 |
false,
|
| 2654 |
false
|
| 2655 |
],
|
|
|
|
| 2698 |
"model.decoder.layers.26.encoder_attn.out_proj.lora_E": [
|
| 2699 |
true,
|
| 2700 |
true,
|
| 2701 |
+
false,
|
| 2702 |
true,
|
| 2703 |
true,
|
| 2704 |
true,
|
|
|
|
| 2706 |
true,
|
| 2707 |
true,
|
| 2708 |
true,
|
| 2709 |
+
false,
|
| 2710 |
true
|
| 2711 |
],
|
| 2712 |
"model.decoder.layers.26.encoder_attn.q_proj.lora_E": [
|
|
|
|
| 2766 |
false
|
| 2767 |
],
|
| 2768 |
"model.decoder.layers.26.self_attn.k_proj.lora_E": [
|
| 2769 |
+
false,
|
| 2770 |
true,
|
| 2771 |
false,
|
| 2772 |
true,
|
|
|
|
| 2774 |
false,
|
| 2775 |
false,
|
| 2776 |
false,
|
| 2777 |
+
false,
|
| 2778 |
true,
|
| 2779 |
false,
|
| 2780 |
false
|
| 2781 |
],
|
| 2782 |
"model.decoder.layers.26.self_attn.out_proj.lora_E": [
|
|
|
|
| 2783 |
false,
|
| 2784 |
false,
|
| 2785 |
true,
|
|
|
|
| 2786 |
true,
|
| 2787 |
+
false,
|
| 2788 |
+
false,
|
| 2789 |
true,
|
| 2790 |
false,
|
| 2791 |
false,
|
|
|
|
| 2794 |
true
|
| 2795 |
],
|
| 2796 |
"model.decoder.layers.26.self_attn.q_proj.lora_E": [
|
|
|
|
| 2797 |
false,
|
|
|
|
| 2798 |
false,
|
| 2799 |
false,
|
| 2800 |
false,
|
|
|
|
| 2803 |
false,
|
| 2804 |
false,
|
| 2805 |
false,
|
| 2806 |
+
false,
|
| 2807 |
+
false,
|
| 2808 |
+
false
|
| 2809 |
],
|
| 2810 |
"model.decoder.layers.26.self_attn.v_proj.lora_E": [
|
| 2811 |
false,
|
| 2812 |
true,
|
| 2813 |
false,
|
|
|
|
|
|
|
| 2814 |
false,
|
| 2815 |
+
true,
|
| 2816 |
false,
|
| 2817 |
false,
|
| 2818 |
true,
|
| 2819 |
true,
|
| 2820 |
+
true,
|
| 2821 |
false,
|
| 2822 |
true
|
| 2823 |
],
|
|
|
|
| 2845 |
true,
|
| 2846 |
true,
|
| 2847 |
true,
|
| 2848 |
+
false,
|
| 2849 |
false,
|
| 2850 |
true
|
| 2851 |
],
|
|
|
|
| 2907 |
],
|
| 2908 |
"model.decoder.layers.27.self_attn.k_proj.lora_E": [
|
| 2909 |
true,
|
| 2910 |
+
false,
|
| 2911 |
true,
|
| 2912 |
false,
|
| 2913 |
false,
|
|
|
|
| 2948 |
false
|
| 2949 |
],
|
| 2950 |
"model.decoder.layers.27.self_attn.v_proj.lora_E": [
|
| 2951 |
+
false,
|
| 2952 |
true,
|
| 2953 |
false,
|
| 2954 |
false,
|
|
|
|
| 2977 |
],
|
| 2978 |
"model.decoder.layers.28.encoder_attn.out_proj.lora_E": [
|
| 2979 |
true,
|
| 2980 |
+
false,
|
| 2981 |
true,
|
| 2982 |
true,
|
| 2983 |
true,
|
| 2984 |
true,
|
|
|
|
|
|
|
| 2985 |
true,
|
| 2986 |
+
false,
|
| 2987 |
true,
|
| 2988 |
true,
|
| 2989 |
true,
|
|
|
|
| 3041 |
true,
|
| 3042 |
true,
|
| 3043 |
true,
|
| 3044 |
+
false,
|
| 3045 |
true,
|
| 3046 |
true
|
| 3047 |
],
|
| 3048 |
"model.decoder.layers.28.self_attn.k_proj.lora_E": [
|
| 3049 |
+
true,
|
| 3050 |
true,
|
| 3051 |
true,
|
| 3052 |
false,
|
|
|
|
| 3068 |
true,
|
| 3069 |
true,
|
| 3070 |
false,
|
| 3071 |
+
false,
|
| 3072 |
false,
|
| 3073 |
false,
|
| 3074 |
false
|
|
|
|
| 3097 |
false,
|
| 3098 |
false,
|
| 3099 |
false,
|
| 3100 |
+
false,
|
| 3101 |
false,
|
| 3102 |
true
|
| 3103 |
],
|
|
|
|
| 3116 |
false
|
| 3117 |
],
|
| 3118 |
"model.decoder.layers.29.encoder_attn.out_proj.lora_E": [
|
|
|
|
|
|
|
| 3119 |
true,
|
| 3120 |
false,
|
| 3121 |
false,
|
| 3122 |
false,
|
|
|
|
| 3123 |
false,
|
| 3124 |
false,
|
| 3125 |
false,
|
| 3126 |
false,
|
| 3127 |
+
false,
|
| 3128 |
+
false,
|
| 3129 |
+
false,
|
| 3130 |
+
false
|
| 3131 |
],
|
| 3132 |
"model.decoder.layers.29.encoder_attn.q_proj.lora_E": [
|
| 3133 |
false,
|
|
|
|
| 3191 |
false,
|
| 3192 |
false,
|
| 3193 |
false,
|
| 3194 |
+
false,
|
| 3195 |
true,
|
| 3196 |
false,
|
| 3197 |
false,
|
|
|
|
| 3210 |
false,
|
| 3211 |
false,
|
| 3212 |
false,
|
| 3213 |
+
false,
|
| 3214 |
false
|
| 3215 |
],
|
| 3216 |
"model.decoder.layers.29.self_attn.q_proj.lora_E": [
|
|
|
|
| 3259 |
true,
|
| 3260 |
true,
|
| 3261 |
false,
|
|
|
|
|
|
|
| 3262 |
false,
|
| 3263 |
true,
|
| 3264 |
+
false,
|
| 3265 |
true,
|
| 3266 |
true,
|
| 3267 |
false,
|
| 3268 |
+
false,
|
| 3269 |
true,
|
| 3270 |
false
|
| 3271 |
],
|
|
|
|
| 3368 |
false
|
| 3369 |
],
|
| 3370 |
"model.decoder.layers.3.self_attn.v_proj.lora_E": [
|
| 3371 |
+
true,
|
| 3372 |
false,
|
| 3373 |
false,
|
| 3374 |
false,
|
|
|
|
| 3378 |
false,
|
| 3379 |
false,
|
| 3380 |
true,
|
| 3381 |
+
false,
|
| 3382 |
false
|
| 3383 |
],
|
| 3384 |
"model.decoder.layers.30.encoder_attn.k_proj.lora_E": [
|
|
|
|
| 3396 |
false
|
| 3397 |
],
|
| 3398 |
"model.decoder.layers.30.encoder_attn.out_proj.lora_E": [
|
|
|
|
| 3399 |
true,
|
| 3400 |
false,
|
| 3401 |
false,
|
| 3402 |
false,
|
| 3403 |
false,
|
| 3404 |
+
false,
|
| 3405 |
+
false,
|
| 3406 |
+
false,
|
| 3407 |
+
false,
|
| 3408 |
true,
|
| 3409 |
+
false,
|
|
|
|
|
|
|
|
|
|
| 3410 |
false
|
| 3411 |
],
|
| 3412 |
"model.decoder.layers.30.encoder_attn.q_proj.lora_E": [
|
|
|
|
| 3475 |
false,
|
| 3476 |
true,
|
| 3477 |
false,
|
| 3478 |
+
true,
|
| 3479 |
true,
|
| 3480 |
true
|
| 3481 |
],
|
|
|
|
| 3517 |
false,
|
| 3518 |
false,
|
| 3519 |
false,
|
| 3520 |
+
false,
|
| 3521 |
false,
|
| 3522 |
true
|
| 3523 |
],
|
|
|
|
| 3584 |
false,
|
| 3585 |
false,
|
| 3586 |
false,
|
| 3587 |
+
false,
|
| 3588 |
false,
|
| 3589 |
true,
|
| 3590 |
true,
|
|
|
|
| 3686 |
false,
|
| 3687 |
false,
|
| 3688 |
false,
|
| 3689 |
+
false,
|
| 3690 |
false
|
| 3691 |
],
|
| 3692 |
"model.decoder.layers.4.encoder_attn.q_proj.lora_E": [
|
|
|
|
| 3705 |
],
|
| 3706 |
"model.decoder.layers.4.encoder_attn.v_proj.lora_E": [
|
| 3707 |
true,
|
| 3708 |
+
false,
|
| 3709 |
false,
|
| 3710 |
false,
|
| 3711 |
false,
|
|
|
|
| 3714 |
true,
|
| 3715 |
false,
|
| 3716 |
true,
|
| 3717 |
+
false,
|
| 3718 |
+
false
|
| 3719 |
],
|
| 3720 |
"model.decoder.layers.4.fc1.lora_E": [
|
| 3721 |
true,
|
|
|
|
| 3853 |
false,
|
| 3854 |
true,
|
| 3855 |
true,
|
| 3856 |
+
false,
|
| 3857 |
false,
|
| 3858 |
true
|
| 3859 |
],
|
|
|
|
| 4048 |
false,
|
| 4049 |
false,
|
| 4050 |
false,
|
| 4051 |
+
false,
|
| 4052 |
false,
|
| 4053 |
false,
|
| 4054 |
false
|
|
|
|
| 4069 |
],
|
| 4070 |
"model.decoder.layers.6.self_attn.v_proj.lora_E": [
|
| 4071 |
false,
|
| 4072 |
+
true,
|
| 4073 |
false,
|
| 4074 |
false,
|
| 4075 |
false,
|
|
|
|
| 4079 |
false,
|
| 4080 |
true,
|
| 4081 |
false,
|
| 4082 |
+
false
|
| 4083 |
],
|
| 4084 |
"model.decoder.layers.7.encoder_attn.k_proj.lora_E": [
|
| 4085 |
false,
|
|
|
|
| 4149 |
true,
|
| 4150 |
true,
|
| 4151 |
true,
|
| 4152 |
+
false
|
| 4153 |
],
|
| 4154 |
"model.decoder.layers.7.fc2.lora_E": [
|
| 4155 |
true,
|
|
|
|
| 4428 |
true,
|
| 4429 |
true,
|
| 4430 |
true,
|
| 4431 |
+
true,
|
| 4432 |
true
|
| 4433 |
],
|
| 4434 |
"model.decoder.layers.9.fc2.lora_E": [
|
|
|
|
| 4443 |
true,
|
| 4444 |
true,
|
| 4445 |
true,
|
| 4446 |
+
false
|
| 4447 |
],
|
| 4448 |
"model.decoder.layers.9.self_attn.k_proj.lora_E": [
|
| 4449 |
false,
|
|
|
|
| 4496 |
true,
|
| 4497 |
false,
|
| 4498 |
false,
|
|
|
|
| 4499 |
false,
|
| 4500 |
false,
|
| 4501 |
+
true,
|
| 4502 |
false
|
| 4503 |
],
|
| 4504 |
"model.encoder.layers.0.fc1.lora_E": [
|
|
|
|
| 4550 |
true,
|
| 4551 |
false,
|
| 4552 |
false,
|
| 4553 |
+
false,
|
| 4554 |
+
false,
|
| 4555 |
false,
|
| 4556 |
false,
|
| 4557 |
false,
|
|
|
|
| 4778 |
false,
|
| 4779 |
true,
|
| 4780 |
false,
|
| 4781 |
+
true,
|
| 4782 |
false
|
| 4783 |
],
|
| 4784 |
"model.encoder.layers.11.self_attn.k_proj.lora_E": [
|
|
|
|
| 4832 |
false,
|
| 4833 |
false,
|
| 4834 |
false,
|
| 4835 |
+
true,
|
| 4836 |
false,
|
| 4837 |
false,
|
| 4838 |
false
|
|
|
|
| 4885 |
false,
|
| 4886 |
false,
|
| 4887 |
false,
|
| 4888 |
+
true,
|
| 4889 |
false,
|
| 4890 |
false,
|
| 4891 |
false,
|
|
|
|
| 4926 |
false,
|
| 4927 |
false,
|
| 4928 |
false,
|
| 4929 |
+
true,
|
| 4930 |
true,
|
| 4931 |
false,
|
| 4932 |
false,
|
|
|
|
| 4943 |
true,
|
| 4944 |
false,
|
| 4945 |
false,
|
| 4946 |
+
true,
|
| 4947 |
false,
|
| 4948 |
false,
|
| 4949 |
true,
|
|
|
|
| 4967 |
false,
|
| 4968 |
false,
|
| 4969 |
false,
|
| 4970 |
+
true,
|
| 4971 |
false,
|
| 4972 |
false,
|
| 4973 |
false,
|
|
|
|
| 5003 |
false,
|
| 5004 |
false,
|
| 5005 |
false,
|
| 5006 |
+
false
|
| 5007 |
],
|
| 5008 |
"model.encoder.layers.14.fc1.lora_E": [
|
| 5009 |
true,
|
|
|
|
| 5016 |
true,
|
| 5017 |
false,
|
| 5018 |
false,
|
| 5019 |
+
false,
|
| 5020 |
true
|
| 5021 |
],
|
| 5022 |
"model.encoder.layers.14.fc2.lora_E": [
|
|
|
|
| 5056 |
false,
|
| 5057 |
false,
|
| 5058 |
true,
|
| 5059 |
+
false,
|
| 5060 |
false,
|
| 5061 |
true,
|
| 5062 |
false
|
|
|
|
| 5136 |
true,
|
| 5137 |
false,
|
| 5138 |
false,
|
| 5139 |
+
false,
|
| 5140 |
true,
|
| 5141 |
false,
|
| 5142 |
false,
|
| 5143 |
+
false,
|
| 5144 |
false,
|
| 5145 |
false,
|
| 5146 |
false
|
|
|
|
| 5168 |
false,
|
| 5169 |
false,
|
| 5170 |
true,
|
| 5171 |
+
false,
|
| 5172 |
false,
|
| 5173 |
false,
|
| 5174 |
true
|
|
|
|
| 5223 |
true,
|
| 5224 |
false,
|
| 5225 |
true,
|
|
|
|
| 5226 |
false,
|
| 5227 |
false,
|
| 5228 |
false,
|
| 5229 |
+
true,
|
| 5230 |
true
|
| 5231 |
],
|
| 5232 |
"model.encoder.layers.16.self_attn.q_proj.lora_E": [
|
|
|
|
| 5307 |
true,
|
| 5308 |
true,
|
| 5309 |
false,
|
| 5310 |
+
false,
|
| 5311 |
true,
|
| 5312 |
+
false,
|
|
|
|
| 5313 |
true,
|
| 5314 |
true
|
| 5315 |
],
|
|
|
|
| 5386 |
"model.encoder.layers.18.self_attn.out_proj.lora_E": [
|
| 5387 |
false,
|
| 5388 |
false,
|
|
|
|
| 5389 |
false,
|
| 5390 |
false,
|
| 5391 |
false,
|
| 5392 |
+
false,
|
| 5393 |
+
false,
|
| 5394 |
+
false,
|
| 5395 |
true,
|
| 5396 |
false,
|
| 5397 |
false,
|
|
|
|
| 5474 |
false,
|
| 5475 |
false,
|
| 5476 |
false,
|
| 5477 |
+
true,
|
| 5478 |
false,
|
| 5479 |
false,
|
| 5480 |
false,
|
|
|
|
| 5806 |
"model.encoder.layers.22.self_attn.out_proj.lora_E": [
|
| 5807 |
false,
|
| 5808 |
false,
|
| 5809 |
+
false,
|
| 5810 |
false,
|
| 5811 |
false,
|
| 5812 |
true,
|
| 5813 |
+
false,
|
| 5814 |
+
false,
|
| 5815 |
true,
|
| 5816 |
false,
|
| 5817 |
false,
|
|
|
|
| 5892 |
true,
|
| 5893 |
true,
|
| 5894 |
true,
|
| 5895 |
+
false,
|
| 5896 |
+
false,
|
| 5897 |
true,
|
| 5898 |
true,
|
| 5899 |
false,
|
| 5900 |
false,
|
| 5901 |
+
false,
|
| 5902 |
true
|
| 5903 |
],
|
| 5904 |
"model.encoder.layers.23.self_attn.q_proj.lora_E": [
|
|
|
|
| 5916 |
false
|
| 5917 |
],
|
| 5918 |
"model.encoder.layers.23.self_attn.v_proj.lora_E": [
|
|
|
|
| 5919 |
true,
|
| 5920 |
false,
|
| 5921 |
false,
|
| 5922 |
+
false,
|
| 5923 |
true,
|
| 5924 |
true,
|
| 5925 |
true,
|
|
|
|
| 6011 |
true,
|
| 6012 |
true,
|
| 6013 |
true,
|
| 6014 |
+
false
|
| 6015 |
],
|
| 6016 |
"model.encoder.layers.25.fc1.lora_E": [
|
| 6017 |
true,
|
|
|
|
| 6092 |
true,
|
| 6093 |
true,
|
| 6094 |
true,
|
| 6095 |
+
false,
|
| 6096 |
true,
|
| 6097 |
true,
|
| 6098 |
true
|
|
|
|
| 6227 |
true,
|
| 6228 |
true,
|
| 6229 |
false,
|
|
|
|
|
|
|
| 6230 |
true,
|
| 6231 |
+
false,
|
| 6232 |
true,
|
| 6233 |
true,
|
| 6234 |
+
false,
|
| 6235 |
true,
|
| 6236 |
false,
|
| 6237 |
false,
|
|
|
|
| 6309 |
],
|
| 6310 |
"model.encoder.layers.28.self_attn.out_proj.lora_E": [
|
| 6311 |
false,
|
| 6312 |
+
false,
|
| 6313 |
false,
|
| 6314 |
true,
|
| 6315 |
false,
|
|
|
|
| 6319 |
true,
|
| 6320 |
true,
|
| 6321 |
true,
|
| 6322 |
+
false
|
| 6323 |
],
|
| 6324 |
"model.encoder.layers.28.self_attn.q_proj.lora_E": [
|
| 6325 |
false,
|
|
|
|
| 6400 |
true,
|
| 6401 |
true,
|
| 6402 |
true,
|
| 6403 |
+
false,
|
| 6404 |
true,
|
| 6405 |
true,
|
| 6406 |
true
|
|
|
|
| 6425 |
true,
|
| 6426 |
true,
|
| 6427 |
true,
|
| 6428 |
+
false,
|
| 6429 |
true,
|
| 6430 |
true,
|
| 6431 |
true,
|
|
|
|
| 6564 |
true,
|
| 6565 |
true,
|
| 6566 |
true,
|
|
|
|
| 6567 |
true,
|
| 6568 |
true,
|
| 6569 |
true,
|
| 6570 |
true,
|
| 6571 |
true,
|
| 6572 |
+
false,
|
| 6573 |
true,
|
| 6574 |
false
|
| 6575 |
],
|
|
|
|
| 6649 |
false,
|
| 6650 |
false,
|
| 6651 |
true,
|
| 6652 |
+
false,
|
| 6653 |
true,
|
| 6654 |
true,
|
| 6655 |
true,
|
|
|
|
| 6677 |
true,
|
| 6678 |
true,
|
| 6679 |
false,
|
| 6680 |
+
false,
|
| 6681 |
false,
|
| 6682 |
true,
|
| 6683 |
true,
|
|
|
|
| 6794 |
false,
|
| 6795 |
false,
|
| 6796 |
false,
|
| 6797 |
+
false,
|
| 6798 |
false
|
| 6799 |
],
|
| 6800 |
"model.encoder.layers.5.self_attn.k_proj.lora_E": [
|
|
|
|
| 6871 |
false,
|
| 6872 |
false,
|
| 6873 |
false,
|
| 6874 |
+
true,
|
| 6875 |
false,
|
| 6876 |
false,
|
| 6877 |
false,
|
|
|
|
| 7040 |
false,
|
| 7041 |
false,
|
| 7042 |
false,
|
| 7043 |
+
false,
|
| 7044 |
false,
|
| 7045 |
false,
|
| 7046 |
false,
|
|
|
|
| 7129 |
false,
|
| 7130 |
false,
|
| 7131 |
false,
|
| 7132 |
+
false,
|
| 7133 |
+
false,
|
| 7134 |
false
|
| 7135 |
],
|
| 7136 |
"model.encoder.layers.9.self_attn.k_proj.lora_E": [
|
best_checkpoint/adapter_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6d8e7c4342456bf184a1699ceee8b79d4637919e86043c74de3fca982e51df19
|
| 3 |
+
size 38697637
|
best_checkpoint/optimizer.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 174313245
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4f14365a17f23ef7c12f92f6afd4d80275de570ff0a27d16f77b510fcb80dec6
|
| 3 |
size 174313245
|
best_checkpoint/random_states_0.pkl
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15691
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:46700cb6c6f878730d63a567eac5fae05f37e7908256976b6e8f7afcfd8cb6d8
|
| 3 |
size 15691
|
best_checkpoint/random_states_1.pkl
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15755
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1ce290220813dc1fc7fecc91c868657014fa05803fefba4b691a6e114ab59d43
|
| 3 |
size 15755
|
best_checkpoint/scheduler.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 563
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6bb58f2d06c0b6169e26724c3f525aff558b10a892707ea045a33661020fdf75
|
| 3 |
size 563
|