Hugging Face's logo

rbelanec
/
test

Text Generation
PEFT
Safetensors
Transformers
llama-factory
conversational
Model card Files
xet
Community
rbelanec commited on
Commit
b1352a6
·
verified ·
1 Parent(s): e677762

Training in progress, step 7

Browse files
Files changed (4) hide show
  1. adapter_config.json +4 -4
  2. adapter_model.safetensors +1 -1
  3. trainer_log.jsonl +14 -43
  4. training_args.bin +1 -1
adapter_config.json CHANGED
@@ -5,12 +5,12 @@
5
  "peft_type": "BITFIT",
6
  "revision": null,
7
  "target_modules": [
8
- "o_proj",
 
9
  "down_proj",
 
10
  "k_proj",
11
- "q_proj",
12
- "up_proj",
13
- "v_proj"
14
  ],
15
  "task_type": "CAUSAL_LM"
16
  }
 
5
  "peft_type": "BITFIT",
6
  "revision": null,
7
  "target_modules": [
8
+ "up_proj",
9
+ "v_proj",
10
  "down_proj",
11
+ "o_proj",
12
  "k_proj",
13
+ "q_proj"
 
 
14
  ],
15
  "task_type": "CAUSAL_LM"
16
  }
adapter_model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:bfbad3b3bbe0780e94ac83258a352ada4e15b8231133708e191bf48a16ed199f
3
  size 2819582528
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4389bc853785f6f0b7a549fab1b3063262a707c18b546f54d132629472ce6fa5
3
  size 2819582528
trainer_log.jsonl CHANGED
@@ -1,43 +1,14 @@
1
- {"current_steps": 5, "total_steps": 125, "loss": 10.9709, "lr": 1.5384615384615387e-05, "epoch": 0.04, "percentage": 4.0, "elapsed_time": "0:00:01", "remaining_time": "0:00:45", "throughput": 1126.49, "total_tokens": 2144}
2
- {"current_steps": 7, "total_steps": 125, "eval_loss": 6.5227251052856445, "epoch": 0.056, "percentage": 5.6, "elapsed_time": "0:00:02", "remaining_time": "0:00:49", "throughput": 985.42, "total_tokens": 2880}
3
- {"current_steps": 10, "total_steps": 125, "loss": 6.4075, "lr": 3.461538461538462e-05, "epoch": 0.08, "percentage": 8.0, "elapsed_time": "0:00:23", "remaining_time": "0:04:25", "throughput": 178.62, "total_tokens": 4128}
4
- {"current_steps": 14, "total_steps": 125, "eval_loss": 1.382468581199646, "epoch": 0.112, "percentage": 11.2, "elapsed_time": "0:00:24", "remaining_time": "0:03:13", "throughput": 242.05, "total_tokens": 5920}
5
- {"current_steps": 15, "total_steps": 125, "loss": 2.5338, "lr": 4.999016565957633e-05, "epoch": 0.12, "percentage": 12.0, "elapsed_time": "0:00:35", "remaining_time": "0:04:21", "throughput": 174.88, "total_tokens": 6240}
6
- {"current_steps": 20, "total_steps": 125, "loss": 0.5326, "lr": 4.96467754629559e-05, "epoch": 0.16, "percentage": 16.0, "elapsed_time": "0:00:36", "remaining_time": "0:03:11", "throughput": 222.01, "total_tokens": 8096}
7
- {"current_steps": 21, "total_steps": 125, "eval_loss": 0.4987373352050781, "epoch": 0.168, "percentage": 16.8, "elapsed_time": "0:00:37", "remaining_time": "0:03:04", "throughput": 225.31, "total_tokens": 8416}
8
- {"current_steps": 25, "total_steps": 125, "loss": 0.4144, "lr": 4.881937806807241e-05, "epoch": 0.2, "percentage": 20.0, "elapsed_time": "0:00:49", "remaining_time": "0:03:17", "throughput": 205.26, "total_tokens": 10112}
9
- {"current_steps": 28, "total_steps": 125, "eval_loss": 0.4531269073486328, "epoch": 0.224, "percentage": 22.4, "elapsed_time": "0:00:50", "remaining_time": "0:02:55", "throughput": 222.87, "total_tokens": 11264}
10
- {"current_steps": 30, "total_steps": 125, "loss": 0.4563, "lr": 4.752422169756048e-05, "epoch": 0.24, "percentage": 24.0, "elapsed_time": "0:01:02", "remaining_time": "0:03:16", "throughput": 193.76, "total_tokens": 12032}
11
- {"current_steps": 35, "total_steps": 125, "loss": 0.4802, "lr": 4.5786740307563636e-05, "epoch": 0.28, "percentage": 28.0, "elapsed_time": "0:01:02", "remaining_time": "0:02:41", "throughput": 219.77, "total_tokens": 13824}
12
- {"current_steps": 35, "total_steps": 125, "eval_loss": 0.36931881308555603, "epoch": 0.28, "percentage": 28.0, "elapsed_time": "0:01:03", "remaining_time": "0:02:44", "throughput": 216.75, "total_tokens": 13824}
13
- {"current_steps": 40, "total_steps": 125, "loss": 0.3809, "lr": 4.364105412207914e-05, "epoch": 0.32, "percentage": 32.0, "elapsed_time": "0:01:14", "remaining_time": "0:02:38", "throughput": 212.26, "total_tokens": 15840}
14
- {"current_steps": 42, "total_steps": 125, "eval_loss": 0.3872639238834381, "epoch": 0.336, "percentage": 33.6, "elapsed_time": "0:01:15", "remaining_time": "0:02:29", "throughput": 220.32, "total_tokens": 16672}
15
- {"current_steps": 45, "total_steps": 125, "loss": 0.3844, "lr": 4.1129299588552193e-05, "epoch": 0.36, "percentage": 36.0, "elapsed_time": "0:01:34", "remaining_time": "0:02:47", "throughput": 190.21, "total_tokens": 17920}
16
- {"current_steps": 49, "total_steps": 125, "eval_loss": 0.3777945637702942, "epoch": 0.392, "percentage": 39.2, "elapsed_time": "0:01:35", "remaining_time": "0:02:28", "throughput": 201.79, "total_tokens": 19296}
17
- {"current_steps": 50, "total_steps": 125, "loss": 0.2817, "lr": 3.830080191288342e-05, "epoch": 0.4, "percentage": 40.0, "elapsed_time": "0:01:46", "remaining_time": "0:02:39", "throughput": 185.1, "total_tokens": 19712}
18
- {"current_steps": 55, "total_steps": 125, "loss": 0.3831, "lr": 3.521110642339991e-05, "epoch": 0.44, "percentage": 44.0, "elapsed_time": "0:01:47", "remaining_time": "0:02:16", "throughput": 204.49, "total_tokens": 21952}
19
- {"current_steps": 56, "total_steps": 125, "eval_loss": 0.4436803460121155, "epoch": 0.448, "percentage": 44.8, "elapsed_time": "0:01:48", "remaining_time": "0:02:13", "throughput": 207.14, "total_tokens": 22432}
20
- {"current_steps": 60, "total_steps": 125, "loss": 0.5576, "lr": 3.1920887785621235e-05, "epoch": 0.48, "percentage": 48.0, "elapsed_time": "0:01:59", "remaining_time": "0:02:09", "throughput": 201.61, "total_tokens": 24160}
21
- {"current_steps": 63, "total_steps": 125, "eval_loss": 0.35032057762145996, "epoch": 0.504, "percentage": 50.4, "elapsed_time": "0:02:01", "remaining_time": "0:01:59", "throughput": 210.33, "total_tokens": 25504}
22
- {"current_steps": 65, "total_steps": 125, "loss": 0.4013, "lr": 2.849475848838749e-05, "epoch": 0.52, "percentage": 52.0, "elapsed_time": "0:02:12", "remaining_time": "0:02:01", "throughput": 197.7, "total_tokens": 26112}
23
- {"current_steps": 70, "total_steps": 125, "loss": 0.3242, "lr": 2.5e-05, "epoch": 0.56, "percentage": 56.0, "elapsed_time": "0:02:12", "remaining_time": "0:01:44", "throughput": 211.13, "total_tokens": 28064}
24
- {"current_steps": 70, "total_steps": 125, "eval_loss": 0.37164703011512756, "epoch": 0.56, "percentage": 56.0, "elapsed_time": "0:02:13", "remaining_time": "0:01:45", "throughput": 209.84, "total_tokens": 28064}
25
- {"current_steps": 75, "total_steps": 125, "loss": 0.3963, "lr": 2.1505241511612522e-05, "epoch": 0.6, "percentage": 60.0, "elapsed_time": "0:02:32", "remaining_time": "0:01:41", "throughput": 195.25, "total_tokens": 29824}
26
- {"current_steps": 77, "total_steps": 125, "eval_loss": 0.3748786747455597, "epoch": 0.616, "percentage": 61.6, "elapsed_time": "0:02:33", "remaining_time": "0:01:35", "throughput": 199.61, "total_tokens": 30720}
27
- {"current_steps": 80, "total_steps": 125, "loss": 0.3946, "lr": 1.8079112214378768e-05, "epoch": 0.64, "percentage": 64.0, "elapsed_time": "0:02:44", "remaining_time": "0:01:32", "throughput": 193.61, "total_tokens": 31904}
28
- {"current_steps": 84, "total_steps": 125, "eval_loss": 0.3603578209877014, "epoch": 0.672, "percentage": 67.2, "elapsed_time": "0:02:46", "remaining_time": "0:01:21", "throughput": 201.5, "total_tokens": 33504}
29
- {"current_steps": 85, "total_steps": 125, "loss": 0.3496, "lr": 1.4788893576600099e-05, "epoch": 0.68, "percentage": 68.0, "elapsed_time": "0:02:54", "remaining_time": "0:01:22", "throughput": 194.65, "total_tokens": 33984}
30
- {"current_steps": 90, "total_steps": 125, "loss": 0.337, "lr": 1.1699198087116589e-05, "epoch": 0.72, "percentage": 72.0, "elapsed_time": "0:02:55", "remaining_time": "0:01:08", "throughput": 203.97, "total_tokens": 35776}
31
- {"current_steps": 91, "total_steps": 125, "eval_loss": 0.35710158944129944, "epoch": 0.728, "percentage": 72.8, "elapsed_time": "0:02:56", "remaining_time": "0:01:05", "throughput": 204.78, "total_tokens": 36128}
32
- {"current_steps": 95, "total_steps": 125, "loss": 0.4315, "lr": 8.870700411447816e-06, "epoch": 0.76, "percentage": 76.0, "elapsed_time": "0:03:05", "remaining_time": "0:00:58", "throughput": 201.87, "total_tokens": 37472}
33
- {"current_steps": 98, "total_steps": 125, "eval_loss": 0.3520326614379883, "epoch": 0.784, "percentage": 78.4, "elapsed_time": "0:03:06", "remaining_time": "0:00:51", "throughput": 206.43, "total_tokens": 38592}
34
- {"current_steps": 100, "total_steps": 125, "loss": 0.38, "lr": 6.358945877920861e-06, "epoch": 0.8, "percentage": 80.0, "elapsed_time": "0:03:25", "remaining_time": "0:00:51", "throughput": 191.58, "total_tokens": 39328}
35
- {"current_steps": 105, "total_steps": 125, "loss": 0.371, "lr": 4.213259692436367e-06, "epoch": 0.84, "percentage": 84.0, "elapsed_time": "0:03:26", "remaining_time": "0:00:39", "throughput": 200.3, "total_tokens": 41280}
36
- {"current_steps": 105, "total_steps": 125, "eval_loss": 0.34758228063583374, "epoch": 0.84, "percentage": 84.0, "elapsed_time": "0:03:26", "remaining_time": "0:00:39", "throughput": 199.52, "total_tokens": 41280}
37
- {"current_steps": 110, "total_steps": 125, "loss": 0.364, "lr": 2.475778302439524e-06, "epoch": 0.88, "percentage": 88.0, "elapsed_time": "0:03:38", "remaining_time": "0:00:29", "throughput": 199.22, "total_tokens": 43552}
38
- {"current_steps": 112, "total_steps": 125, "eval_loss": 0.34589245915412903, "epoch": 0.896, "percentage": 89.6, "elapsed_time": "0:03:39", "remaining_time": "0:00:25", "throughput": 200.95, "total_tokens": 44160}
39
- {"current_steps": 115, "total_steps": 125, "loss": 0.3554, "lr": 1.180621931927592e-06, "epoch": 0.92, "percentage": 92.0, "elapsed_time": "0:03:47", "remaining_time": "0:00:19", "throughput": 198.32, "total_tokens": 45216}
40
- {"current_steps": 119, "total_steps": 125, "eval_loss": 0.3492301404476166, "epoch": 0.952, "percentage": 95.2, "elapsed_time": "0:03:49", "remaining_time": "0:00:11", "throughput": 204.57, "total_tokens": 46944}
41
- {"current_steps": 120, "total_steps": 125, "loss": 0.3494, "lr": 3.5322453704410286e-07, "epoch": 0.96, "percentage": 96.0, "elapsed_time": "0:03:56", "remaining_time": "0:00:09", "throughput": 200.16, "total_tokens": 47360}
42
- {"current_steps": 125, "total_steps": 125, "loss": 0.3588, "lr": 9.834340423678368e-09, "epoch": 1.0, "percentage": 100.0, "elapsed_time": "0:03:57", "remaining_time": "0:00:00", "throughput": 207.81, "total_tokens": 49376}
43
- {"current_steps": 125, "total_steps": 125, "epoch": 1.0, "percentage": 100.0, "elapsed_time": "0:04:22", "remaining_time": "0:00:00", "throughput": 188.19, "total_tokens": 49376}
 
1
+ {"current_steps": 5, "total_steps": 125, "loss": 10.9709, "lr": 1.5384615384615387e-05, "epoch": 0.04, "percentage": 4.0, "elapsed_time": "0:00:02", "remaining_time": "0:01:09", "throughput": 735.72, "total_tokens": 2144}
2
+ {"current_steps": 7, "total_steps": 125, "eval_loss": 6.5227251052856445, "epoch": 0.056, "percentage": 5.6, "elapsed_time": "0:00:03", "remaining_time": "0:01:06", "throughput": 732.42, "total_tokens": 2880}
3
+ {"current_steps": 10, "total_steps": 125, "loss": 6.4075, "lr": 3.461538461538462e-05, "epoch": 0.08, "percentage": 8.0, "elapsed_time": "0:00:20", "remaining_time": "0:03:52", "throughput": 203.87, "total_tokens": 4128}
4
+ {"current_steps": 14, "total_steps": 125, "eval_loss": 1.382468581199646, "epoch": 0.112, "percentage": 11.2, "elapsed_time": "0:00:21", "remaining_time": "0:02:51", "throughput": 273.75, "total_tokens": 5920}
5
+ {"current_steps": 15, "total_steps": 125, "loss": 2.5338, "lr": 4.999016565957633e-05, "epoch": 0.12, "percentage": 12.0, "elapsed_time": "0:00:30", "remaining_time": "0:03:43", "throughput": 204.83, "total_tokens": 6240}
6
+ {"current_steps": 20, "total_steps": 125, "loss": 0.5326, "lr": 4.96467754629559e-05, "epoch": 0.16, "percentage": 16.0, "elapsed_time": "0:00:31", "remaining_time": "0:02:44", "throughput": 259.14, "total_tokens": 8096}
7
+ {"current_steps": 21, "total_steps": 125, "eval_loss": 0.4987373352050781, "epoch": 0.168, "percentage": 16.8, "elapsed_time": "0:00:31", "remaining_time": "0:02:38", "throughput": 263.74, "total_tokens": 8416}
8
+ {"current_steps": 25, "total_steps": 125, "loss": 0.4144, "lr": 4.881937806807241e-05, "epoch": 0.2, "percentage": 20.0, "elapsed_time": "0:00:39", "remaining_time": "0:02:37", "throughput": 256.96, "total_tokens": 10112}
9
+ {"current_steps": 28, "total_steps": 125, "eval_loss": 0.4531269073486328, "epoch": 0.224, "percentage": 22.4, "elapsed_time": "0:00:40", "remaining_time": "0:02:19", "throughput": 279.63, "total_tokens": 11264}
10
+ {"current_steps": 30, "total_steps": 125, "loss": 0.4563, "lr": 4.752422169756048e-05, "epoch": 0.24, "percentage": 24.0, "elapsed_time": "0:00:47", "remaining_time": "0:02:30", "throughput": 253.43, "total_tokens": 12032}
11
+ {"current_steps": 35, "total_steps": 125, "loss": 0.4802, "lr": 4.5786740307563636e-05, "epoch": 0.28, "percentage": 28.0, "elapsed_time": "0:00:48", "remaining_time": "0:02:04", "throughput": 286.57, "total_tokens": 13824}
12
+ {"current_steps": 35, "total_steps": 125, "eval_loss": 0.36931881308555603, "epoch": 0.28, "percentage": 28.0, "elapsed_time": "0:00:48", "remaining_time": "0:02:05", "throughput": 282.37, "total_tokens": 13824}
13
+ {"current_steps": 40, "total_steps": 125, "loss": 0.3809, "lr": 4.364105412207914e-05, "epoch": 0.32, "percentage": 32.0, "elapsed_time": "0:00:58", "remaining_time": "0:02:05", "throughput": 268.48, "total_tokens": 15840}
14
+ {"current_steps": 42, "total_steps": 125, "eval_loss": 0.3872639238834381, "epoch": 0.336, "percentage": 33.6, "elapsed_time": "0:01:00", "remaining_time": "0:01:58", "throughput": 277.51, "total_tokens": 16672}
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:2748056f7e661a88a39d01dcfeb31641675efffc05f00f667e18bb847f69af83
3
  size 6097
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:acc568c9cbb33de9a3eafbda48735873409500bc6aefe53fe5eab764d8b19ad5
3
  size 6097