Hugging Face's logo

sailplane
/
vm-llama3-70b-lr4e-5-be5e-4-r8-dist

Safetensors
Model card Files Community
vgaraujov commited on
Commit
c24d806
·
verified ·
1 Parent(s): 2acce9e

Training in progress, epoch 4, checkpoint

Browse files
Files changed (13) hide show
  1. checkpoint-408/adapter_config.json +3 -3
  2. checkpoint-408/adapter_model.safetensors +1 -1
  3. checkpoint-408/global_step408/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt +1 -1
  4. checkpoint-408/global_step408/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt +1 -1
  5. checkpoint-408/global_step408/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt +1 -1
  6. checkpoint-408/global_step408/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt +1 -1
  7. checkpoint-408/global_step408/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt +1 -1
  8. checkpoint-408/global_step408/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt +1 -1
  9. checkpoint-408/global_step408/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt +1 -1
  10. checkpoint-408/global_step408/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt +1 -1
  11. checkpoint-408/trainer_state.json +5 -5
  12. checkpoint-408/training_args.bin +1 -1
  13. checkpoint-408/value_head.safetensors +1 -1
checkpoint-408/adapter_config.json CHANGED
@@ -20,12 +20,12 @@
20
  "rank_pattern": {},
21
  "revision": null,
22
  "target_modules": [
23
- "down_proj",
24
  "k_proj",
25
- "up_proj",
26
  "o_proj",
 
27
  "q_proj",
28
- "v_proj",
29
  "gate_proj"
30
  ],
31
  "task_type": "CAUSAL_LM",
 
20
  "rank_pattern": {},
21
  "revision": null,
22
  "target_modules": [
23
+ "v_proj",
24
  "k_proj",
25
+ "down_proj",
26
  "o_proj",
27
+ "up_proj",
28
  "q_proj",
 
29
  "gate_proj"
30
  ],
31
  "task_type": "CAUSAL_LM",
checkpoint-408/adapter_model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:9c51b8500522dee31551811e4b16a900fb9a021c2985a322c9ee6a2057fdbf83
3
  size 207244392
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ab2047dcb9b3de00092860b228d5147a5cf15f2bf3f8411d1a631348abd320c5
3
  size 207244392
checkpoint-408/global_step408/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:0771f18bc959c215a1c471d98092500b5dc937277c6511c5c3225393d86a54a3
3
  size 19421232
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cbf73ceec9b61a4ddbfbdbc6636363652d496b087ef1bec85dbd23c2be0ab91f
3
  size 19421232
checkpoint-408/global_step408/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:399f3bd539ac06ddde2809c58a7e4b5a5d9008b137596a3bdb9dba406c4aa5b6
3
  size 19421232
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3dc2c695bb0e5496692434af0c44b9d519dbe100be7c01238c3e9047bd44026c
3
  size 19421232
checkpoint-408/global_step408/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:a91de3b433c52adc3a3e17c3a1013b59c40a249b344dad787f6850cca6a4e885
3
  size 19421232
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:526b693fe11933a1b6b2d3ad8be002944280953f629cf21b0c79683752c265b4
3
  size 19421232
checkpoint-408/global_step408/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:a6b8cbedae73c5003c3ad9c0d717d8e561030a49b8370ab0e3fd7c06aca3c4d3
3
  size 19421232
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:63a115a1b56ddbaf44273edf6bfce316f82d0a1ec2f8e552f5f3923eabf4e09f
3
  size 19421232
checkpoint-408/global_step408/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:aeccf28474fb7672906b095a459833ee051b943a749676952f6677d127b21ae3
3
  size 19421232
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7b58cb1e750d5d5fc7425b241cd8efb9b777c7196b9a1192857f2369a4c4ccba
3
  size 19421232
checkpoint-408/global_step408/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:7aceb033251f722253298ab7f75d272a2e4bed53c78cebd75d3d04e55e15f3c0
3
  size 19421232
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:59afa6c8d157fac420f4da196a2872b6923505542ce54802de2bf573306e3361
3
  size 19421232
checkpoint-408/global_step408/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:cb204bab8b5e225e1934b1b3f7d374d76ec6cc11f354eb4595033f700b1fc84c
3
  size 19421232
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0a2f572eb37063a3bb191e311d94ac6c34ed5658187953a47335638473c529bd
3
  size 19421232
checkpoint-408/global_step408/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:cb1e44b4ef63fdf90467f47980b82c72deb1cc2ab4c5cbb4a87158ad2cbceaec
3
  size 19421232
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cade71fcf086a8f535eeb7db994610e40bba589f4e09effd4ada4a5030fbe8c1
3
  size 19421232
checkpoint-408/trainer_state.json CHANGED
@@ -10,28 +10,28 @@
10
  "log_history": [
11
  {
12
  "epoch": 0.9803921568627451,
13
- "grad_norm": 0.04069041460752487,
14
  "learning_rate": 3.9722621383652286e-05,
15
  "loss": 0.3224,
16
  "step": 100
17
  },
18
  {
19
  "epoch": 1.9607843137254903,
20
- "grad_norm": 0.06058349087834358,
21
  "learning_rate": 3.816756982694638e-05,
22
- "loss": 0.2133,
23
  "step": 200
24
  },
25
  {
26
  "epoch": 2.9411764705882355,
27
- "grad_norm": 0.07736460864543915,
28
  "learning_rate": 3.534732018082765e-05,
29
  "loss": 0.1972,
30
  "step": 300
31
  },
32
  {
33
  "epoch": 3.9215686274509802,
34
- "grad_norm": 0.11341462284326553,
35
  "learning_rate": 3.145827595429179e-05,
36
  "loss": 0.1842,
37
  "step": 400
 
10
  "log_history": [
11
  {
12
  "epoch": 0.9803921568627451,
13
+ "grad_norm": 0.041187554597854614,
14
  "learning_rate": 3.9722621383652286e-05,
15
  "loss": 0.3224,
16
  "step": 100
17
  },
18
  {
19
  "epoch": 1.9607843137254903,
20
+ "grad_norm": 0.05797708407044411,
21
  "learning_rate": 3.816756982694638e-05,
22
+ "loss": 0.2134,
23
  "step": 200
24
  },
25
  {
26
  "epoch": 2.9411764705882355,
27
+ "grad_norm": 0.07534682005643845,
28
  "learning_rate": 3.534732018082765e-05,
29
  "loss": 0.1972,
30
  "step": 300
31
  },
32
  {
33
  "epoch": 3.9215686274509802,
34
+ "grad_norm": 0.11367125064134598,
35
  "learning_rate": 3.145827595429179e-05,
36
  "loss": 0.1842,
37
  "step": 400
checkpoint-408/training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:007bcc1d400d4a233d19c941cf50d51f13137550797eb8625a06e27dc172670e
3
  size 7352
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e85267171c46dc1c3303c31870ee78028f6f93ce2789cc5b73fc96c18746b9e2
3
  size 7352
checkpoint-408/value_head.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:6ab6967a0b476bdb45b037b4d7704a54eefb484ed61053528af9ba66b0239dfa
3
  size 16586
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:53c8a581f25de66ced51ce80fa3adca98627380c44243e80e94099626bba00b5
3
  size 16586