tktung commited on
Commit
ba9cd26
·
verified ·
1 Parent(s): be45956

Upload folder using huggingface_hub

Browse files
Files changed (20) hide show
  1. uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/config.json +30 -0
  2. uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/generation_config.json +10 -0
  3. uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/latest +1 -0
  4. uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/model-00001-of-00006.safetensors +3 -0
  5. uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/model-00002-of-00006.safetensors +3 -0
  6. uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/model-00003-of-00006.safetensors +3 -0
  7. uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/model-00004-of-00006.safetensors +3 -0
  8. uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/model-00005-of-00006.safetensors +3 -0
  9. uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/model-00006-of-00006.safetensors +3 -0
  10. uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/model.safetensors.index.json +370 -0
  11. uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/rng_state_0.pth +3 -0
  12. uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/rng_state_1.pth +3 -0
  13. uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/scheduler.pt +3 -0
  14. uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/special_tokens_map.json +24 -0
  15. uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/tokenizer.json +0 -0
  16. uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/tokenizer.model +3 -0
  17. uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/tokenizer_config.json +43 -0
  18. uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/trainer_state.json +593 -0
  19. uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/training_args.bin +3 -0
  20. uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/zero_to_fp32.py +674 -0
uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/config.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "/mnt/data/tungtran/output_model/irish_llama2_data_v3/checkpoint-2200",
3
+ "architectures": [
4
+ "LlamaForCausalLM"
5
+ ],
6
+ "attention_bias": false,
7
+ "attention_dropout": 0.0,
8
+ "bos_token_id": 1,
9
+ "eos_token_id": 2,
10
+ "head_dim": 128,
11
+ "hidden_act": "silu",
12
+ "hidden_size": 5120,
13
+ "initializer_range": 0.02,
14
+ "intermediate_size": 13824,
15
+ "max_position_embeddings": 4096,
16
+ "mlp_bias": false,
17
+ "model_type": "llama",
18
+ "num_attention_heads": 40,
19
+ "num_hidden_layers": 40,
20
+ "num_key_value_heads": 40,
21
+ "pretraining_tp": 1,
22
+ "rms_norm_eps": 1e-05,
23
+ "rope_scaling": null,
24
+ "rope_theta": 10000.0,
25
+ "tie_word_embeddings": false,
26
+ "torch_dtype": "bfloat16",
27
+ "transformers_version": "4.46.3",
28
+ "use_cache": true,
29
+ "vocab_size": 35483
30
+ }
uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/generation_config.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token_id": 1,
3
+ "do_sample": true,
4
+ "eos_token_id": 2,
5
+ "max_length": 4096,
6
+ "pad_token_id": 0,
7
+ "temperature": 0.6,
8
+ "top_p": 0.9,
9
+ "transformers_version": "4.46.3"
10
+ }
uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/latest ADDED
@@ -0,0 +1 @@
 
 
1
+ global_step158
uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/model-00001-of-00006.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d3d6c3b211eae68b6dabb734cc6fbf1d26d0095212f38ae7964b79dd0de92270
3
+ size 4961502800
uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/model-00002-of-00006.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cf2ff969c9ec88ea045767aaa36475a11c61108a0ef53d6af58f79a1e434dc22
3
+ size 4970422232
uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/model-00003-of-00006.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:618aa18fa5283797cec1c51faf0dc223f55599b55b9937bb347925911bc530f1
3
+ size 4881272584
uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/model-00004-of-00006.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7d62dde4967c7cca1f2072e48afdd4b9f480bc022fa80d484e8e96b126876e16
3
+ size 4933722216
uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/model-00005-of-00006.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b324943a05d25d051630fd1536c7520d262197925773536f0925ac07aba27179
3
+ size 4933722208
uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/model-00006-of-00006.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:269e550ba87cdeee01ba568fd44a5dc8d8f062d5811f200338b04afb58e4bc59
3
+ size 1422460712
uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/model.safetensors.index.json ADDED
@@ -0,0 +1,370 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_size": 26103060480
4
+ },
5
+ "weight_map": {
6
+ "lm_head.weight": "model-00006-of-00006.safetensors",
7
+ "model.embed_tokens.weight": "model-00001-of-00006.safetensors",
8
+ "model.layers.0.input_layernorm.weight": "model-00001-of-00006.safetensors",
9
+ "model.layers.0.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
10
+ "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00006.safetensors",
11
+ "model.layers.0.mlp.up_proj.weight": "model-00001-of-00006.safetensors",
12
+ "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00006.safetensors",
13
+ "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
14
+ "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
15
+ "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
16
+ "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
17
+ "model.layers.1.input_layernorm.weight": "model-00001-of-00006.safetensors",
18
+ "model.layers.1.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
19
+ "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00006.safetensors",
20
+ "model.layers.1.mlp.up_proj.weight": "model-00001-of-00006.safetensors",
21
+ "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00006.safetensors",
22
+ "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
23
+ "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
24
+ "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
25
+ "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
26
+ "model.layers.10.input_layernorm.weight": "model-00002-of-00006.safetensors",
27
+ "model.layers.10.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
28
+ "model.layers.10.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
29
+ "model.layers.10.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
30
+ "model.layers.10.post_attention_layernorm.weight": "model-00002-of-00006.safetensors",
31
+ "model.layers.10.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
32
+ "model.layers.10.self_attn.o_proj.weight": "model-00002-of-00006.safetensors",
33
+ "model.layers.10.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
34
+ "model.layers.10.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
35
+ "model.layers.11.input_layernorm.weight": "model-00002-of-00006.safetensors",
36
+ "model.layers.11.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
37
+ "model.layers.11.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
38
+ "model.layers.11.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
39
+ "model.layers.11.post_attention_layernorm.weight": "model-00002-of-00006.safetensors",
40
+ "model.layers.11.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
41
+ "model.layers.11.self_attn.o_proj.weight": "model-00002-of-00006.safetensors",
42
+ "model.layers.11.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
43
+ "model.layers.11.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
44
+ "model.layers.12.input_layernorm.weight": "model-00002-of-00006.safetensors",
45
+ "model.layers.12.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
46
+ "model.layers.12.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
47
+ "model.layers.12.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
48
+ "model.layers.12.post_attention_layernorm.weight": "model-00002-of-00006.safetensors",
49
+ "model.layers.12.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
50
+ "model.layers.12.self_attn.o_proj.weight": "model-00002-of-00006.safetensors",
51
+ "model.layers.12.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
52
+ "model.layers.12.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
53
+ "model.layers.13.input_layernorm.weight": "model-00002-of-00006.safetensors",
54
+ "model.layers.13.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
55
+ "model.layers.13.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
56
+ "model.layers.13.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
57
+ "model.layers.13.post_attention_layernorm.weight": "model-00002-of-00006.safetensors",
58
+ "model.layers.13.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
59
+ "model.layers.13.self_attn.o_proj.weight": "model-00002-of-00006.safetensors",
60
+ "model.layers.13.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
61
+ "model.layers.13.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
62
+ "model.layers.14.input_layernorm.weight": "model-00002-of-00006.safetensors",
63
+ "model.layers.14.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
64
+ "model.layers.14.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
65
+ "model.layers.14.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
66
+ "model.layers.14.post_attention_layernorm.weight": "model-00002-of-00006.safetensors",
67
+ "model.layers.14.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
68
+ "model.layers.14.self_attn.o_proj.weight": "model-00002-of-00006.safetensors",
69
+ "model.layers.14.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
70
+ "model.layers.14.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
71
+ "model.layers.15.input_layernorm.weight": "model-00003-of-00006.safetensors",
72
+ "model.layers.15.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
73
+ "model.layers.15.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
74
+ "model.layers.15.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
75
+ "model.layers.15.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
76
+ "model.layers.15.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
77
+ "model.layers.15.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
78
+ "model.layers.15.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
79
+ "model.layers.15.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
80
+ "model.layers.16.input_layernorm.weight": "model-00003-of-00006.safetensors",
81
+ "model.layers.16.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
82
+ "model.layers.16.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
83
+ "model.layers.16.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
84
+ "model.layers.16.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
85
+ "model.layers.16.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
86
+ "model.layers.16.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
87
+ "model.layers.16.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
88
+ "model.layers.16.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
89
+ "model.layers.17.input_layernorm.weight": "model-00003-of-00006.safetensors",
90
+ "model.layers.17.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
91
+ "model.layers.17.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
92
+ "model.layers.17.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
93
+ "model.layers.17.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
94
+ "model.layers.17.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
95
+ "model.layers.17.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
96
+ "model.layers.17.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
97
+ "model.layers.17.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
98
+ "model.layers.18.input_layernorm.weight": "model-00003-of-00006.safetensors",
99
+ "model.layers.18.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
100
+ "model.layers.18.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
101
+ "model.layers.18.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
102
+ "model.layers.18.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
103
+ "model.layers.18.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
104
+ "model.layers.18.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
105
+ "model.layers.18.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
106
+ "model.layers.18.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
107
+ "model.layers.19.input_layernorm.weight": "model-00003-of-00006.safetensors",
108
+ "model.layers.19.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
109
+ "model.layers.19.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
110
+ "model.layers.19.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
111
+ "model.layers.19.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
112
+ "model.layers.19.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
113
+ "model.layers.19.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
114
+ "model.layers.19.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
115
+ "model.layers.19.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
116
+ "model.layers.2.input_layernorm.weight": "model-00001-of-00006.safetensors",
117
+ "model.layers.2.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
118
+ "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00006.safetensors",
119
+ "model.layers.2.mlp.up_proj.weight": "model-00001-of-00006.safetensors",
120
+ "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00006.safetensors",
121
+ "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
122
+ "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
123
+ "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
124
+ "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
125
+ "model.layers.20.input_layernorm.weight": "model-00003-of-00006.safetensors",
126
+ "model.layers.20.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
127
+ "model.layers.20.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
128
+ "model.layers.20.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
129
+ "model.layers.20.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
130
+ "model.layers.20.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
131
+ "model.layers.20.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
132
+ "model.layers.20.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
133
+ "model.layers.20.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
134
+ "model.layers.21.input_layernorm.weight": "model-00003-of-00006.safetensors",
135
+ "model.layers.21.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
136
+ "model.layers.21.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
137
+ "model.layers.21.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
138
+ "model.layers.21.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
139
+ "model.layers.21.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
140
+ "model.layers.21.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
141
+ "model.layers.21.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
142
+ "model.layers.21.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
143
+ "model.layers.22.input_layernorm.weight": "model-00004-of-00006.safetensors",
144
+ "model.layers.22.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
145
+ "model.layers.22.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
146
+ "model.layers.22.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
147
+ "model.layers.22.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
148
+ "model.layers.22.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
149
+ "model.layers.22.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
150
+ "model.layers.22.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
151
+ "model.layers.22.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
152
+ "model.layers.23.input_layernorm.weight": "model-00004-of-00006.safetensors",
153
+ "model.layers.23.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
154
+ "model.layers.23.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
155
+ "model.layers.23.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
156
+ "model.layers.23.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
157
+ "model.layers.23.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
158
+ "model.layers.23.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
159
+ "model.layers.23.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
160
+ "model.layers.23.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
161
+ "model.layers.24.input_layernorm.weight": "model-00004-of-00006.safetensors",
162
+ "model.layers.24.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
163
+ "model.layers.24.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
164
+ "model.layers.24.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
165
+ "model.layers.24.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
166
+ "model.layers.24.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
167
+ "model.layers.24.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
168
+ "model.layers.24.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
169
+ "model.layers.24.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
170
+ "model.layers.25.input_layernorm.weight": "model-00004-of-00006.safetensors",
171
+ "model.layers.25.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
172
+ "model.layers.25.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
173
+ "model.layers.25.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
174
+ "model.layers.25.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
175
+ "model.layers.25.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
176
+ "model.layers.25.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
177
+ "model.layers.25.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
178
+ "model.layers.25.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
179
+ "model.layers.26.input_layernorm.weight": "model-00004-of-00006.safetensors",
180
+ "model.layers.26.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
181
+ "model.layers.26.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
182
+ "model.layers.26.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
183
+ "model.layers.26.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
184
+ "model.layers.26.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
185
+ "model.layers.26.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
186
+ "model.layers.26.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
187
+ "model.layers.26.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
188
+ "model.layers.27.input_layernorm.weight": "model-00004-of-00006.safetensors",
189
+ "model.layers.27.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
190
+ "model.layers.27.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
191
+ "model.layers.27.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
192
+ "model.layers.27.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
193
+ "model.layers.27.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
194
+ "model.layers.27.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
195
+ "model.layers.27.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
196
+ "model.layers.27.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
197
+ "model.layers.28.input_layernorm.weight": "model-00004-of-00006.safetensors",
198
+ "model.layers.28.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
199
+ "model.layers.28.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
200
+ "model.layers.28.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
201
+ "model.layers.28.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
202
+ "model.layers.28.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
203
+ "model.layers.28.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
204
+ "model.layers.28.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
205
+ "model.layers.28.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
206
+ "model.layers.29.input_layernorm.weight": "model-00004-of-00006.safetensors",
207
+ "model.layers.29.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
208
+ "model.layers.29.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
209
+ "model.layers.29.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
210
+ "model.layers.29.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
211
+ "model.layers.29.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
212
+ "model.layers.29.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
213
+ "model.layers.29.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
214
+ "model.layers.29.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
215
+ "model.layers.3.input_layernorm.weight": "model-00001-of-00006.safetensors",
216
+ "model.layers.3.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
217
+ "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00006.safetensors",
218
+ "model.layers.3.mlp.up_proj.weight": "model-00001-of-00006.safetensors",
219
+ "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00006.safetensors",
220
+ "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
221
+ "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
222
+ "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
223
+ "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
224
+ "model.layers.30.input_layernorm.weight": "model-00005-of-00006.safetensors",
225
+ "model.layers.30.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
226
+ "model.layers.30.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
227
+ "model.layers.30.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
228
+ "model.layers.30.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
229
+ "model.layers.30.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
230
+ "model.layers.30.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
231
+ "model.layers.30.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
232
+ "model.layers.30.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
233
+ "model.layers.31.input_layernorm.weight": "model-00005-of-00006.safetensors",
234
+ "model.layers.31.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
235
+ "model.layers.31.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
236
+ "model.layers.31.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
237
+ "model.layers.31.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
238
+ "model.layers.31.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
239
+ "model.layers.31.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
240
+ "model.layers.31.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
241
+ "model.layers.31.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
242
+ "model.layers.32.input_layernorm.weight": "model-00005-of-00006.safetensors",
243
+ "model.layers.32.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
244
+ "model.layers.32.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
245
+ "model.layers.32.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
246
+ "model.layers.32.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
247
+ "model.layers.32.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
248
+ "model.layers.32.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
249
+ "model.layers.32.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
250
+ "model.layers.32.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
251
+ "model.layers.33.input_layernorm.weight": "model-00005-of-00006.safetensors",
252
+ "model.layers.33.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
253
+ "model.layers.33.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
254
+ "model.layers.33.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
255
+ "model.layers.33.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
256
+ "model.layers.33.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
257
+ "model.layers.33.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
258
+ "model.layers.33.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
259
+ "model.layers.33.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
260
+ "model.layers.34.input_layernorm.weight": "model-00005-of-00006.safetensors",
261
+ "model.layers.34.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
262
+ "model.layers.34.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
263
+ "model.layers.34.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
264
+ "model.layers.34.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
265
+ "model.layers.34.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
266
+ "model.layers.34.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
267
+ "model.layers.34.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
268
+ "model.layers.34.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
269
+ "model.layers.35.input_layernorm.weight": "model-00005-of-00006.safetensors",
270
+ "model.layers.35.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
271
+ "model.layers.35.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
272
+ "model.layers.35.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
273
+ "model.layers.35.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
274
+ "model.layers.35.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
275
+ "model.layers.35.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
276
+ "model.layers.35.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
277
+ "model.layers.35.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
278
+ "model.layers.36.input_layernorm.weight": "model-00005-of-00006.safetensors",
279
+ "model.layers.36.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
280
+ "model.layers.36.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
281
+ "model.layers.36.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
282
+ "model.layers.36.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
283
+ "model.layers.36.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
284
+ "model.layers.36.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
285
+ "model.layers.36.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
286
+ "model.layers.36.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
287
+ "model.layers.37.input_layernorm.weight": "model-00005-of-00006.safetensors",
288
+ "model.layers.37.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
289
+ "model.layers.37.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
290
+ "model.layers.37.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
291
+ "model.layers.37.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
292
+ "model.layers.37.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
293
+ "model.layers.37.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
294
+ "model.layers.37.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
295
+ "model.layers.37.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
296
+ "model.layers.38.input_layernorm.weight": "model-00006-of-00006.safetensors",
297
+ "model.layers.38.mlp.down_proj.weight": "model-00006-of-00006.safetensors",
298
+ "model.layers.38.mlp.gate_proj.weight": "model-00006-of-00006.safetensors",
299
+ "model.layers.38.mlp.up_proj.weight": "model-00006-of-00006.safetensors",
300
+ "model.layers.38.post_attention_layernorm.weight": "model-00006-of-00006.safetensors",
301
+ "model.layers.38.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
302
+ "model.layers.38.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
303
+ "model.layers.38.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
304
+ "model.layers.38.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
305
+ "model.layers.39.input_layernorm.weight": "model-00006-of-00006.safetensors",
306
+ "model.layers.39.mlp.down_proj.weight": "model-00006-of-00006.safetensors",
307
+ "model.layers.39.mlp.gate_proj.weight": "model-00006-of-00006.safetensors",
308
+ "model.layers.39.mlp.up_proj.weight": "model-00006-of-00006.safetensors",
309
+ "model.layers.39.post_attention_layernorm.weight": "model-00006-of-00006.safetensors",
310
+ "model.layers.39.self_attn.k_proj.weight": "model-00006-of-00006.safetensors",
311
+ "model.layers.39.self_attn.o_proj.weight": "model-00006-of-00006.safetensors",
312
+ "model.layers.39.self_attn.q_proj.weight": "model-00006-of-00006.safetensors",
313
+ "model.layers.39.self_attn.v_proj.weight": "model-00006-of-00006.safetensors",
314
+ "model.layers.4.input_layernorm.weight": "model-00001-of-00006.safetensors",
315
+ "model.layers.4.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
316
+ "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00006.safetensors",
317
+ "model.layers.4.mlp.up_proj.weight": "model-00001-of-00006.safetensors",
318
+ "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00006.safetensors",
319
+ "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
320
+ "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
321
+ "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
322
+ "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
323
+ "model.layers.5.input_layernorm.weight": "model-00001-of-00006.safetensors",
324
+ "model.layers.5.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
325
+ "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00006.safetensors",
326
+ "model.layers.5.mlp.up_proj.weight": "model-00001-of-00006.safetensors",
327
+ "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00006.safetensors",
328
+ "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
329
+ "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
330
+ "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
331
+ "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
332
+ "model.layers.6.input_layernorm.weight": "model-00001-of-00006.safetensors",
333
+ "model.layers.6.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
334
+ "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00006.safetensors",
335
+ "model.layers.6.mlp.up_proj.weight": "model-00001-of-00006.safetensors",
336
+ "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00006.safetensors",
337
+ "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
338
+ "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
339
+ "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
340
+ "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
341
+ "model.layers.7.input_layernorm.weight": "model-00002-of-00006.safetensors",
342
+ "model.layers.7.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
343
+ "model.layers.7.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
344
+ "model.layers.7.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
345
+ "model.layers.7.post_attention_layernorm.weight": "model-00002-of-00006.safetensors",
346
+ "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
347
+ "model.layers.7.self_attn.o_proj.weight": "model-00002-of-00006.safetensors",
348
+ "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
349
+ "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
350
+ "model.layers.8.input_layernorm.weight": "model-00002-of-00006.safetensors",
351
+ "model.layers.8.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
352
+ "model.layers.8.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
353
+ "model.layers.8.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
354
+ "model.layers.8.post_attention_layernorm.weight": "model-00002-of-00006.safetensors",
355
+ "model.layers.8.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
356
+ "model.layers.8.self_attn.o_proj.weight": "model-00002-of-00006.safetensors",
357
+ "model.layers.8.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
358
+ "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
359
+ "model.layers.9.input_layernorm.weight": "model-00002-of-00006.safetensors",
360
+ "model.layers.9.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
361
+ "model.layers.9.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
362
+ "model.layers.9.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
363
+ "model.layers.9.post_attention_layernorm.weight": "model-00002-of-00006.safetensors",
364
+ "model.layers.9.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
365
+ "model.layers.9.self_attn.o_proj.weight": "model-00002-of-00006.safetensors",
366
+ "model.layers.9.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
367
+ "model.layers.9.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
368
+ "model.norm.weight": "model-00006-of-00006.safetensors"
369
+ }
370
+ }
uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b365b711bf2b8c15c57134e1c074de625587243159e0f28ad3b8daec4accff67
3
+ size 14512
uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7a8a055a9dfd6455398509d9bd165bcaf91e9f2f7302609c0b932b5447428d7e
3
+ size 14512
uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6654bf60c4bad0ada943fd8fce7b7de172ed06d975375149a6b2062b31cea6ea
3
+ size 1064
uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "</s>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3d1f5d0342153f3e3bbb37b2026ba64d0b25583df351345f87cd8b9a5658c2fb
3
+ size 558602
uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/tokenizer_config.json ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "add_prefix_space": true,
5
+ "added_tokens_decoder": {
6
+ "0": {
7
+ "content": "<unk>",
8
+ "lstrip": false,
9
+ "normalized": false,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": false,
26
+ "rstrip": false,
27
+ "single_word": false,
28
+ "special": true
29
+ }
30
+ },
31
+ "bos_token": "<s>",
32
+ "chat_template": "{% if messages[0]['role'] == 'system' %}{% set loop_messages = messages[1:] %}{% set system_message = messages[0]['content'] %}{% else %}{% set loop_messages = messages %}{% set system_message = 'You are a helpful, respectful and honest assistant. Always answer as helpfully as possible, while being safe.' %}{% endif %}{% for message in loop_messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if loop.index0 == 0 and system_message != false %}{% set content = '<<SYS>>\\n' + system_message + '\\n<</SYS>>\\n\\n' + message['content'] %}{% else %}{% set content = message['content'] %}{% endif %}{% if message['role'] == 'user' %}{{ bos_token + '[INST] ' + content.strip() + ' [/INST]' }}{% elif message['role'] == 'assistant' %}{{ ' ' + content.strip() + ' ' + eos_token }}{% endif %}{% endfor %}",
33
+ "clean_up_tokenization_spaces": false,
34
+ "eos_token": "</s>",
35
+ "legacy": true,
36
+ "model_max_length": 1000000000000000019884624838656,
37
+ "pad_token": "</s>",
38
+ "sp_model_kwargs": {},
39
+ "spaces_between_special_tokens": false,
40
+ "tokenizer_class": "LlamaTokenizer",
41
+ "unk_token": "<unk>",
42
+ "use_default_system_prompt": false
43
+ }
uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/trainer_state.json ADDED
@@ -0,0 +1,593 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 1.9905511811023622,
5
+ "eval_steps": 500,
6
+ "global_step": 158,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.012598425196850394,
13
+ "grad_norm": 1.130629539489746,
14
+ "learning_rate": 6.25e-06,
15
+ "loss": 1.6965,
16
+ "step": 1
17
+ },
18
+ {
19
+ "epoch": 0.025196850393700787,
20
+ "grad_norm": 1.1764330863952637,
21
+ "learning_rate": 1.25e-05,
22
+ "loss": 1.7042,
23
+ "step": 2
24
+ },
25
+ {
26
+ "epoch": 0.050393700787401574,
27
+ "grad_norm": 0.18122103810310364,
28
+ "learning_rate": 2.5e-05,
29
+ "loss": 1.5295,
30
+ "step": 4
31
+ },
32
+ {
33
+ "epoch": 0.07559055118110236,
34
+ "grad_norm": 0.4274924695491791,
35
+ "learning_rate": 3.7500000000000003e-05,
36
+ "loss": 1.3971,
37
+ "step": 6
38
+ },
39
+ {
40
+ "epoch": 0.10078740157480315,
41
+ "grad_norm": 0.7392313480377197,
42
+ "learning_rate": 5e-05,
43
+ "loss": 1.5099,
44
+ "step": 8
45
+ },
46
+ {
47
+ "epoch": 0.12598425196850394,
48
+ "grad_norm": 0.29404416680336,
49
+ "learning_rate": 6.25e-05,
50
+ "loss": 1.4144,
51
+ "step": 10
52
+ },
53
+ {
54
+ "epoch": 0.15118110236220472,
55
+ "grad_norm": 0.18568824231624603,
56
+ "learning_rate": 7.500000000000001e-05,
57
+ "loss": 1.3341,
58
+ "step": 12
59
+ },
60
+ {
61
+ "epoch": 0.1763779527559055,
62
+ "grad_norm": 0.16120545566082,
63
+ "learning_rate": 8.75e-05,
64
+ "loss": 1.3011,
65
+ "step": 14
66
+ },
67
+ {
68
+ "epoch": 0.2015748031496063,
69
+ "grad_norm": 0.09310784935951233,
70
+ "learning_rate": 0.0001,
71
+ "loss": 1.2458,
72
+ "step": 16
73
+ },
74
+ {
75
+ "epoch": 0.22677165354330708,
76
+ "grad_norm": 0.07340509444475174,
77
+ "learning_rate": 9.998903417374228e-05,
78
+ "loss": 1.215,
79
+ "step": 18
80
+ },
81
+ {
82
+ "epoch": 0.25196850393700787,
83
+ "grad_norm": 0.06960318982601166,
84
+ "learning_rate": 9.995614150494293e-05,
85
+ "loss": 1.2052,
86
+ "step": 20
87
+ },
88
+ {
89
+ "epoch": 0.27716535433070866,
90
+ "grad_norm": 0.058640073984861374,
91
+ "learning_rate": 9.990133642141359e-05,
92
+ "loss": 1.1913,
93
+ "step": 22
94
+ },
95
+ {
96
+ "epoch": 0.30236220472440944,
97
+ "grad_norm": 0.04140308499336243,
98
+ "learning_rate": 9.982464296247522e-05,
99
+ "loss": 1.1544,
100
+ "step": 24
101
+ },
102
+ {
103
+ "epoch": 0.32755905511811023,
104
+ "grad_norm": 0.05067910999059677,
105
+ "learning_rate": 9.972609476841367e-05,
106
+ "loss": 1.146,
107
+ "step": 26
108
+ },
109
+ {
110
+ "epoch": 0.352755905511811,
111
+ "grad_norm": 0.03361475095152855,
112
+ "learning_rate": 9.96057350657239e-05,
113
+ "loss": 1.1136,
114
+ "step": 28
115
+ },
116
+ {
117
+ "epoch": 0.3779527559055118,
118
+ "grad_norm": 0.029164660722017288,
119
+ "learning_rate": 9.946361664814943e-05,
120
+ "loss": 1.1128,
121
+ "step": 30
122
+ },
123
+ {
124
+ "epoch": 0.4031496062992126,
125
+ "grad_norm": 0.03163473680615425,
126
+ "learning_rate": 9.929980185352526e-05,
127
+ "loss": 1.0801,
128
+ "step": 32
129
+ },
130
+ {
131
+ "epoch": 0.4283464566929134,
132
+ "grad_norm": 0.02621687948703766,
133
+ "learning_rate": 9.911436253643445e-05,
134
+ "loss": 1.0853,
135
+ "step": 34
136
+ },
137
+ {
138
+ "epoch": 0.45354330708661417,
139
+ "grad_norm": 0.02314259298145771,
140
+ "learning_rate": 9.890738003669029e-05,
141
+ "loss": 1.0942,
142
+ "step": 36
143
+ },
144
+ {
145
+ "epoch": 0.47874015748031495,
146
+ "grad_norm": 0.027407709509134293,
147
+ "learning_rate": 9.867894514365802e-05,
148
+ "loss": 1.074,
149
+ "step": 38
150
+ },
151
+ {
152
+ "epoch": 0.5039370078740157,
153
+ "grad_norm": 0.02250426821410656,
154
+ "learning_rate": 9.842915805643155e-05,
155
+ "loss": 1.0547,
156
+ "step": 40
157
+ },
158
+ {
159
+ "epoch": 0.5291338582677165,
160
+ "grad_norm": 0.023746158927679062,
161
+ "learning_rate": 9.815812833988291e-05,
162
+ "loss": 1.0583,
163
+ "step": 42
164
+ },
165
+ {
166
+ "epoch": 0.5543307086614173,
167
+ "grad_norm": 0.020694848150014877,
168
+ "learning_rate": 9.786597487660337e-05,
169
+ "loss": 1.0374,
170
+ "step": 44
171
+ },
172
+ {
173
+ "epoch": 0.5795275590551181,
174
+ "grad_norm": 0.021605506539344788,
175
+ "learning_rate": 9.755282581475769e-05,
176
+ "loss": 1.0561,
177
+ "step": 46
178
+ },
179
+ {
180
+ "epoch": 0.6047244094488189,
181
+ "grad_norm": 0.08317892998456955,
182
+ "learning_rate": 9.721881851187406e-05,
183
+ "loss": 1.0554,
184
+ "step": 48
185
+ },
186
+ {
187
+ "epoch": 0.6299212598425197,
188
+ "grad_norm": 0.026709511876106262,
189
+ "learning_rate": 9.686409947459458e-05,
190
+ "loss": 1.049,
191
+ "step": 50
192
+ },
193
+ {
194
+ "epoch": 0.6551181102362205,
195
+ "grad_norm": 0.022177977487444878,
196
+ "learning_rate": 9.648882429441257e-05,
197
+ "loss": 1.0254,
198
+ "step": 52
199
+ },
200
+ {
201
+ "epoch": 0.6803149606299213,
202
+ "grad_norm": 0.05779964104294777,
203
+ "learning_rate": 9.609315757942503e-05,
204
+ "loss": 1.0084,
205
+ "step": 54
206
+ },
207
+ {
208
+ "epoch": 0.705511811023622,
209
+ "grad_norm": 0.03699235990643501,
210
+ "learning_rate": 9.567727288213005e-05,
211
+ "loss": 1.0201,
212
+ "step": 56
213
+ },
214
+ {
215
+ "epoch": 0.7307086614173228,
216
+ "grad_norm": 0.02411213330924511,
217
+ "learning_rate": 9.524135262330098e-05,
218
+ "loss": 1.0278,
219
+ "step": 58
220
+ },
221
+ {
222
+ "epoch": 0.7559055118110236,
223
+ "grad_norm": 0.030064478516578674,
224
+ "learning_rate": 9.478558801197065e-05,
225
+ "loss": 1.0288,
226
+ "step": 60
227
+ },
228
+ {
229
+ "epoch": 0.7811023622047244,
230
+ "grad_norm": 0.023079903796315193,
231
+ "learning_rate": 9.431017896156074e-05,
232
+ "loss": 1.0174,
233
+ "step": 62
234
+ },
235
+ {
236
+ "epoch": 0.8062992125984252,
237
+ "grad_norm": 0.025795839726924896,
238
+ "learning_rate": 9.381533400219318e-05,
239
+ "loss": 1.0124,
240
+ "step": 64
241
+ },
242
+ {
243
+ "epoch": 0.831496062992126,
244
+ "grad_norm": 0.022699084132909775,
245
+ "learning_rate": 9.330127018922194e-05,
246
+ "loss": 1.0029,
247
+ "step": 66
248
+ },
249
+ {
250
+ "epoch": 0.8566929133858268,
251
+ "grad_norm": 0.020135775208473206,
252
+ "learning_rate": 9.276821300802534e-05,
253
+ "loss": 1.0141,
254
+ "step": 68
255
+ },
256
+ {
257
+ "epoch": 0.8818897637795275,
258
+ "grad_norm": 0.022621216252446175,
259
+ "learning_rate": 9.221639627510076e-05,
260
+ "loss": 1.0075,
261
+ "step": 70
262
+ },
263
+ {
264
+ "epoch": 0.9070866141732283,
265
+ "grad_norm": 0.020864592865109444,
266
+ "learning_rate": 9.164606203550497e-05,
267
+ "loss": 1.0161,
268
+ "step": 72
269
+ },
270
+ {
271
+ "epoch": 0.9322834645669291,
272
+ "grad_norm": 0.02512463368475437,
273
+ "learning_rate": 9.105746045668521e-05,
274
+ "loss": 1.0012,
275
+ "step": 74
276
+ },
277
+ {
278
+ "epoch": 0.9574803149606299,
279
+ "grad_norm": 0.019894642755389214,
280
+ "learning_rate": 9.045084971874738e-05,
281
+ "loss": 0.9899,
282
+ "step": 76
283
+ },
284
+ {
285
+ "epoch": 0.9826771653543307,
286
+ "grad_norm": 0.025451743975281715,
287
+ "learning_rate": 8.982649590120982e-05,
288
+ "loss": 0.9971,
289
+ "step": 78
290
+ },
291
+ {
292
+ "epoch": 1.0078740157480315,
293
+ "grad_norm": 0.046239905059337616,
294
+ "learning_rate": 8.9184672866292e-05,
295
+ "loss": 1.2662,
296
+ "step": 80
297
+ },
298
+ {
299
+ "epoch": 1.0330708661417323,
300
+ "grad_norm": 0.03749828413128853,
301
+ "learning_rate": 8.852566213878947e-05,
302
+ "loss": 0.8471,
303
+ "step": 82
304
+ },
305
+ {
306
+ "epoch": 1.058267716535433,
307
+ "grad_norm": 0.03041001409292221,
308
+ "learning_rate": 8.784975278258783e-05,
309
+ "loss": 0.846,
310
+ "step": 84
311
+ },
312
+ {
313
+ "epoch": 1.0834645669291338,
314
+ "grad_norm": 0.028458639979362488,
315
+ "learning_rate": 8.715724127386972e-05,
316
+ "loss": 0.8422,
317
+ "step": 86
318
+ },
319
+ {
320
+ "epoch": 1.1086614173228346,
321
+ "grad_norm": 0.06002620980143547,
322
+ "learning_rate": 8.644843137107059e-05,
323
+ "loss": 0.8542,
324
+ "step": 88
325
+ },
326
+ {
327
+ "epoch": 1.1338582677165354,
328
+ "grad_norm": 0.1820724755525589,
329
+ "learning_rate": 8.572363398164017e-05,
330
+ "loss": 0.8322,
331
+ "step": 90
332
+ },
333
+ {
334
+ "epoch": 1.1590551181102362,
335
+ "grad_norm": 0.03150517866015434,
336
+ "learning_rate": 8.498316702566828e-05,
337
+ "loss": 0.8502,
338
+ "step": 92
339
+ },
340
+ {
341
+ "epoch": 1.184251968503937,
342
+ "grad_norm": 0.022805744782090187,
343
+ "learning_rate": 8.422735529643444e-05,
344
+ "loss": 0.8344,
345
+ "step": 94
346
+ },
347
+ {
348
+ "epoch": 1.2094488188976378,
349
+ "grad_norm": 0.04208308458328247,
350
+ "learning_rate": 8.345653031794292e-05,
351
+ "loss": 0.835,
352
+ "step": 96
353
+ },
354
+ {
355
+ "epoch": 1.2346456692913386,
356
+ "grad_norm": 0.025858385488390923,
357
+ "learning_rate": 8.267103019950529e-05,
358
+ "loss": 0.8403,
359
+ "step": 98
360
+ },
361
+ {
362
+ "epoch": 1.2598425196850394,
363
+ "grad_norm": 0.02568437159061432,
364
+ "learning_rate": 8.18711994874345e-05,
365
+ "loss": 0.8322,
366
+ "step": 100
367
+ },
368
+ {
369
+ "epoch": 1.2850393700787401,
370
+ "grad_norm": 0.023638654500246048,
371
+ "learning_rate": 8.105738901391552e-05,
372
+ "loss": 0.8213,
373
+ "step": 102
374
+ },
375
+ {
376
+ "epoch": 1.310236220472441,
377
+ "grad_norm": 0.02360583283007145,
378
+ "learning_rate": 8.022995574311876e-05,
379
+ "loss": 0.8261,
380
+ "step": 104
381
+ },
382
+ {
383
+ "epoch": 1.3354330708661417,
384
+ "grad_norm": 0.02193664200603962,
385
+ "learning_rate": 7.938926261462366e-05,
386
+ "loss": 0.8414,
387
+ "step": 106
388
+ },
389
+ {
390
+ "epoch": 1.3606299212598425,
391
+ "grad_norm": 0.2472202032804489,
392
+ "learning_rate": 7.85356783842216e-05,
393
+ "loss": 0.8178,
394
+ "step": 108
395
+ },
396
+ {
397
+ "epoch": 1.3858267716535433,
398
+ "grad_norm": 0.06591206789016724,
399
+ "learning_rate": 7.766957746216721e-05,
400
+ "loss": 0.8298,
401
+ "step": 110
402
+ },
403
+ {
404
+ "epoch": 1.411023622047244,
405
+ "grad_norm": 0.04800654202699661,
406
+ "learning_rate": 7.679133974894983e-05,
407
+ "loss": 0.8348,
408
+ "step": 112
409
+ },
410
+ {
411
+ "epoch": 1.4362204724409449,
412
+ "grad_norm": 0.028064914047718048,
413
+ "learning_rate": 7.590135046865651e-05,
414
+ "loss": 0.8178,
415
+ "step": 114
416
+ },
417
+ {
418
+ "epoch": 1.4614173228346456,
419
+ "grad_norm": 0.027114123106002808,
420
+ "learning_rate": 7.500000000000001e-05,
421
+ "loss": 0.8141,
422
+ "step": 116
423
+ },
424
+ {
425
+ "epoch": 1.4866141732283464,
426
+ "grad_norm": 0.033917464315891266,
427
+ "learning_rate": 7.408768370508576e-05,
428
+ "loss": 0.8467,
429
+ "step": 118
430
+ },
431
+ {
432
+ "epoch": 1.5118110236220472,
433
+ "grad_norm": 0.026465928182005882,
434
+ "learning_rate": 7.316480175599309e-05,
435
+ "loss": 0.8221,
436
+ "step": 120
437
+ },
438
+ {
439
+ "epoch": 1.537007874015748,
440
+ "grad_norm": 0.027052806690335274,
441
+ "learning_rate": 7.223175895924638e-05,
442
+ "loss": 0.8189,
443
+ "step": 122
444
+ },
445
+ {
446
+ "epoch": 1.5622047244094488,
447
+ "grad_norm": 0.08486244827508926,
448
+ "learning_rate": 7.128896457825364e-05,
449
+ "loss": 0.8062,
450
+ "step": 124
451
+ },
452
+ {
453
+ "epoch": 1.5874015748031496,
454
+ "grad_norm": 0.023071881383657455,
455
+ "learning_rate": 7.033683215379002e-05,
456
+ "loss": 0.8099,
457
+ "step": 126
458
+ },
459
+ {
460
+ "epoch": 1.6125984251968504,
461
+ "grad_norm": 0.02159295789897442,
462
+ "learning_rate": 6.937577932260515e-05,
463
+ "loss": 0.8159,
464
+ "step": 128
465
+ },
466
+ {
467
+ "epoch": 1.6377952755905512,
468
+ "grad_norm": 0.020891694352030754,
469
+ "learning_rate": 6.840622763423391e-05,
470
+ "loss": 0.8076,
471
+ "step": 130
472
+ },
473
+ {
474
+ "epoch": 1.662992125984252,
475
+ "grad_norm": 0.019092196598649025,
476
+ "learning_rate": 6.742860236609077e-05,
477
+ "loss": 0.8276,
478
+ "step": 132
479
+ },
480
+ {
481
+ "epoch": 1.6881889763779527,
482
+ "grad_norm": 0.024381063878536224,
483
+ "learning_rate": 6.644333233692916e-05,
484
+ "loss": 0.8377,
485
+ "step": 134
486
+ },
487
+ {
488
+ "epoch": 1.7133858267716535,
489
+ "grad_norm": 0.020351463928818703,
490
+ "learning_rate": 6.545084971874738e-05,
491
+ "loss": 0.8331,
492
+ "step": 136
493
+ },
494
+ {
495
+ "epoch": 1.7385826771653543,
496
+ "grad_norm": 0.020380325615406036,
497
+ "learning_rate": 6.445158984722358e-05,
498
+ "loss": 0.8158,
499
+ "step": 138
500
+ },
501
+ {
502
+ "epoch": 1.763779527559055,
503
+ "grad_norm": 0.019676068797707558,
504
+ "learning_rate": 6.344599103076329e-05,
505
+ "loss": 0.821,
506
+ "step": 140
507
+ },
508
+ {
509
+ "epoch": 1.7889763779527559,
510
+ "grad_norm": 0.01985323056578636,
511
+ "learning_rate": 6.243449435824276e-05,
512
+ "loss": 0.8237,
513
+ "step": 142
514
+ },
515
+ {
516
+ "epoch": 1.8141732283464567,
517
+ "grad_norm": 0.02249792218208313,
518
+ "learning_rate": 6.141754350553279e-05,
519
+ "loss": 0.8259,
520
+ "step": 144
521
+ },
522
+ {
523
+ "epoch": 1.8393700787401575,
524
+ "grad_norm": 0.020342253148555756,
525
+ "learning_rate": 6.0395584540887963e-05,
526
+ "loss": 0.8117,
527
+ "step": 146
528
+ },
529
+ {
530
+ "epoch": 1.8645669291338582,
531
+ "grad_norm": 0.02552032843232155,
532
+ "learning_rate": 5.9369065729286245e-05,
533
+ "loss": 0.8096,
534
+ "step": 148
535
+ },
536
+ {
537
+ "epoch": 1.889763779527559,
538
+ "grad_norm": 0.019198371097445488,
539
+ "learning_rate": 5.833843733580512e-05,
540
+ "loss": 0.8062,
541
+ "step": 150
542
+ },
543
+ {
544
+ "epoch": 1.9149606299212598,
545
+ "grad_norm": 0.019546708092093468,
546
+ "learning_rate": 5.730415142812059e-05,
547
+ "loss": 0.7951,
548
+ "step": 152
549
+ },
550
+ {
551
+ "epoch": 1.9401574803149606,
552
+ "grad_norm": 0.020258421078324318,
553
+ "learning_rate": 5.6266661678215216e-05,
554
+ "loss": 0.828,
555
+ "step": 154
556
+ },
557
+ {
558
+ "epoch": 1.9653543307086614,
559
+ "grad_norm": 0.023072510957717896,
560
+ "learning_rate": 5.522642316338268e-05,
561
+ "loss": 0.8149,
562
+ "step": 156
563
+ },
564
+ {
565
+ "epoch": 1.9905511811023622,
566
+ "grad_norm": 0.019452733919024467,
567
+ "learning_rate": 5.418389216661579e-05,
568
+ "loss": 0.8239,
569
+ "step": 158
570
+ }
571
+ ],
572
+ "logging_steps": 2,
573
+ "max_steps": 316,
574
+ "num_input_tokens_seen": 0,
575
+ "num_train_epochs": 4,
576
+ "save_steps": 500,
577
+ "stateful_callbacks": {
578
+ "TrainerControl": {
579
+ "args": {
580
+ "should_epoch_stop": false,
581
+ "should_evaluate": false,
582
+ "should_log": false,
583
+ "should_save": true,
584
+ "should_training_stop": false
585
+ },
586
+ "attributes": {}
587
+ }
588
+ },
589
+ "total_flos": 413883181301760.0,
590
+ "train_batch_size": 8,
591
+ "trial_name": null,
592
+ "trial_params": null
593
+ }
uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:637a1faa442a175bf6d4ed7f26d9066abe0385ac2cce7aed134252810cb0b3e1
3
+ size 8056
uccix_v2_instruct_191224_mixture_lr1e-4_fixed_data_mixture/checkpoint-158/zero_to_fp32.py ADDED
@@ -0,0 +1,674 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/usr/bin/env python
2
+
3
+ # Copyright (c) Microsoft Corporation.
4
+ # SPDX-License-Identifier: Apache-2.0
5
+
6
+ # DeepSpeed Team
7
+
8
+ # This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
9
+ # copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
10
+ # the future. Once extracted, the weights don't require DeepSpeed and can be used in any
11
+ # application.
12
+ #
13
+ # example:
14
+ # python zero_to_fp32.py . output_dir/
15
+ # or
16
+ # python zero_to_fp32.py . output_dir/ --safe_serialization
17
+
18
+ import argparse
19
+ import torch
20
+ import glob
21
+ import math
22
+ import os
23
+ import re
24
+ import json
25
+ from tqdm import tqdm
26
+ from collections import OrderedDict
27
+ from dataclasses import dataclass
28
+
29
+ # while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
30
+ # DeepSpeed data structures it has to be available in the current python environment.
31
+ from deepspeed.utils import logger
32
+ from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
33
+ FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
34
+ FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
35
+
36
+
37
+ @dataclass
38
+ class zero_model_state:
39
+ buffers: dict()
40
+ param_shapes: dict()
41
+ shared_params: list
42
+ ds_version: int
43
+ frozen_param_shapes: dict()
44
+ frozen_param_fragments: dict()
45
+
46
+
47
+ debug = 0
48
+
49
+ # load to cpu
50
+ device = torch.device('cpu')
51
+
52
+
53
+ def atoi(text):
54
+ return int(text) if text.isdigit() else text
55
+
56
+
57
+ def natural_keys(text):
58
+ '''
59
+ alist.sort(key=natural_keys) sorts in human order
60
+ http://nedbatchelder.com/blog/200712/human_sorting.html
61
+ (See Toothy's implementation in the comments)
62
+ '''
63
+ return [atoi(c) for c in re.split(r'(\d+)', text)]
64
+
65
+
66
+ def get_model_state_file(checkpoint_dir, zero_stage):
67
+ if not os.path.isdir(checkpoint_dir):
68
+ raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
69
+
70
+ # there should be only one file
71
+ if zero_stage <= 2:
72
+ file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
73
+ elif zero_stage == 3:
74
+ file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
75
+
76
+ if not os.path.exists(file):
77
+ raise FileNotFoundError(f"can't find model states file at '{file}'")
78
+
79
+ return file
80
+
81
+
82
+ def get_checkpoint_files(checkpoint_dir, glob_pattern):
83
+ # XXX: need to test that this simple glob rule works for multi-node setup too
84
+ ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
85
+
86
+ if len(ckpt_files) == 0:
87
+ raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
88
+
89
+ return ckpt_files
90
+
91
+
92
+ def get_optim_files(checkpoint_dir):
93
+ return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
94
+
95
+
96
+ def get_model_state_files(checkpoint_dir):
97
+ return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
98
+
99
+
100
+ def parse_model_states(files):
101
+ zero_model_states = []
102
+ for file in files:
103
+ state_dict = torch.load(file, map_location=device)
104
+
105
+ if BUFFER_NAMES not in state_dict:
106
+ raise ValueError(f"{file} is not a model state checkpoint")
107
+ buffer_names = state_dict[BUFFER_NAMES]
108
+ if debug:
109
+ print("Found buffers:", buffer_names)
110
+
111
+ # recover just the buffers while restoring them to fp32 if they were saved in fp16
112
+ buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
113
+ param_shapes = state_dict[PARAM_SHAPES]
114
+
115
+ # collect parameters that are included in param_shapes
116
+ param_names = []
117
+ for s in param_shapes:
118
+ for name in s.keys():
119
+ param_names.append(name)
120
+
121
+ # update with frozen parameters
122
+ frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
123
+ if frozen_param_shapes is not None:
124
+ if debug:
125
+ print(f"Found frozen_param_shapes: {frozen_param_shapes}")
126
+ param_names += list(frozen_param_shapes.keys())
127
+
128
+ # handle shared params
129
+ shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
130
+
131
+ ds_version = state_dict.get(DS_VERSION, None)
132
+
133
+ frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
134
+
135
+ z_model_state = zero_model_state(buffers=buffers,
136
+ param_shapes=param_shapes,
137
+ shared_params=shared_params,
138
+ ds_version=ds_version,
139
+ frozen_param_shapes=frozen_param_shapes,
140
+ frozen_param_fragments=frozen_param_fragments)
141
+ zero_model_states.append(z_model_state)
142
+
143
+ return zero_model_states
144
+
145
+
146
+ def parse_optim_states(files, ds_checkpoint_dir):
147
+ total_files = len(files)
148
+ state_dicts = []
149
+ for f in files:
150
+ state_dict = torch.load(f, map_location=device)
151
+ # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
152
+ # and also handle the case where it was already removed by another helper script
153
+ state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
154
+ state_dicts.append(state_dict)
155
+
156
+ if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
157
+ raise ValueError(f"{files[0]} is not a zero checkpoint")
158
+ zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
159
+ world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
160
+
161
+ # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
162
+ # parameters can be different from data parallelism for non-expert parameters. So we can just
163
+ # use the max of the partition_count to get the dp world_size.
164
+
165
+ if type(world_size) is list:
166
+ world_size = max(world_size)
167
+
168
+ if world_size != total_files:
169
+ raise ValueError(
170
+ f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
171
+ "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
172
+ )
173
+
174
+ # the groups are named differently in each stage
175
+ if zero_stage <= 2:
176
+ fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
177
+ elif zero_stage == 3:
178
+ fp32_groups_key = FP32_FLAT_GROUPS
179
+ else:
180
+ raise ValueError(f"unknown zero stage {zero_stage}")
181
+
182
+ if zero_stage <= 2:
183
+ fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
184
+ elif zero_stage == 3:
185
+ # if there is more than one param group, there will be multiple flattened tensors - one
186
+ # flattened tensor per group - for simplicity merge them into a single tensor
187
+ #
188
+ # XXX: could make the script more memory efficient for when there are multiple groups - it
189
+ # will require matching the sub-lists of param_shapes for each param group flattened tensor
190
+
191
+ fp32_flat_groups = [
192
+ torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
193
+ ]
194
+
195
+ return zero_stage, world_size, fp32_flat_groups
196
+
197
+
198
+ def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters):
199
+ """
200
+ Returns fp32 state_dict reconstructed from ds checkpoint
201
+
202
+ Args:
203
+ - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
204
+
205
+ """
206
+ print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
207
+
208
+ optim_files = get_optim_files(ds_checkpoint_dir)
209
+ zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
210
+ print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
211
+
212
+ model_files = get_model_state_files(ds_checkpoint_dir)
213
+
214
+ zero_model_states = parse_model_states(model_files)
215
+ print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
216
+
217
+ if zero_stage <= 2:
218
+ return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
219
+ exclude_frozen_parameters)
220
+ elif zero_stage == 3:
221
+ return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
222
+ exclude_frozen_parameters)
223
+
224
+
225
+ def _zero2_merge_frozen_params(state_dict, zero_model_states):
226
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
227
+ return
228
+
229
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
230
+ frozen_param_fragments = zero_model_states[0].frozen_param_fragments
231
+
232
+ if debug:
233
+ num_elem = sum(s.numel() for s in frozen_param_shapes.values())
234
+ print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
235
+
236
+ wanted_params = len(frozen_param_shapes)
237
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
238
+ avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
239
+ print(f'Frozen params: Have {avail_numel} numels to process.')
240
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
241
+
242
+ total_params = 0
243
+ total_numel = 0
244
+ for name, shape in frozen_param_shapes.items():
245
+ total_params += 1
246
+ unpartitioned_numel = shape.numel()
247
+ total_numel += unpartitioned_numel
248
+
249
+ state_dict[name] = frozen_param_fragments[name]
250
+
251
+ if debug:
252
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
253
+
254
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
255
+
256
+
257
+ def _has_callable(obj, fn):
258
+ attr = getattr(obj, fn, None)
259
+ return callable(attr)
260
+
261
+
262
+ def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
263
+ param_shapes = zero_model_states[0].param_shapes
264
+
265
+ # Reconstruction protocol:
266
+ #
267
+ # XXX: document this
268
+
269
+ if debug:
270
+ for i in range(world_size):
271
+ for j in range(len(fp32_flat_groups[0])):
272
+ print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
273
+
274
+ # XXX: memory usage doubles here (zero2)
275
+ num_param_groups = len(fp32_flat_groups[0])
276
+ merged_single_partition_of_fp32_groups = []
277
+ for i in range(num_param_groups):
278
+ merged_partitions = [sd[i] for sd in fp32_flat_groups]
279
+ full_single_fp32_vector = torch.cat(merged_partitions, 0)
280
+ merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
281
+ avail_numel = sum(
282
+ [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
283
+
284
+ if debug:
285
+ wanted_params = sum([len(shapes) for shapes in param_shapes])
286
+ wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
287
+ # not asserting if there is a mismatch due to possible padding
288
+ print(f"Have {avail_numel} numels to process.")
289
+ print(f"Need {wanted_numel} numels in {wanted_params} params.")
290
+
291
+ # params
292
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
293
+ # out-of-core computing solution
294
+ total_numel = 0
295
+ total_params = 0
296
+ for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
297
+ offset = 0
298
+ avail_numel = full_single_fp32_vector.numel()
299
+ for name, shape in shapes.items():
300
+
301
+ unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
302
+ total_numel += unpartitioned_numel
303
+ total_params += 1
304
+
305
+ if debug:
306
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
307
+ state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
308
+ offset += unpartitioned_numel
309
+
310
+ # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
311
+ # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
312
+ # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
313
+ # live optimizer object, so we are checking that the numbers are within the right range
314
+ align_to = 2 * world_size
315
+
316
+ def zero2_align(x):
317
+ return align_to * math.ceil(x / align_to)
318
+
319
+ if debug:
320
+ print(f"original offset={offset}, avail_numel={avail_numel}")
321
+
322
+ offset = zero2_align(offset)
323
+ avail_numel = zero2_align(avail_numel)
324
+
325
+ if debug:
326
+ print(f"aligned offset={offset}, avail_numel={avail_numel}")
327
+
328
+ # Sanity check
329
+ if offset != avail_numel:
330
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
331
+
332
+ print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
333
+
334
+
335
+ def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
336
+ exclude_frozen_parameters):
337
+ state_dict = OrderedDict()
338
+
339
+ # buffers
340
+ buffers = zero_model_states[0].buffers
341
+ state_dict.update(buffers)
342
+ if debug:
343
+ print(f"added {len(buffers)} buffers")
344
+
345
+ if not exclude_frozen_parameters:
346
+ _zero2_merge_frozen_params(state_dict, zero_model_states)
347
+
348
+ _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
349
+
350
+ # recover shared parameters
351
+ for pair in zero_model_states[0].shared_params:
352
+ if pair[1] in state_dict:
353
+ state_dict[pair[0]] = state_dict[pair[1]]
354
+
355
+ return state_dict
356
+
357
+
358
+ def zero3_partitioned_param_info(unpartitioned_numel, world_size):
359
+ remainder = unpartitioned_numel % world_size
360
+ padding_numel = (world_size - remainder) if remainder else 0
361
+ partitioned_numel = math.ceil(unpartitioned_numel / world_size)
362
+ return partitioned_numel, padding_numel
363
+
364
+
365
+ def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
366
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
367
+ return
368
+
369
+ if debug:
370
+ for i in range(world_size):
371
+ num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
372
+ print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
373
+
374
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
375
+ wanted_params = len(frozen_param_shapes)
376
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
377
+ avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
378
+ print(f'Frozen params: Have {avail_numel} numels to process.')
379
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
380
+
381
+ total_params = 0
382
+ total_numel = 0
383
+ for name, shape in zero_model_states[0].frozen_param_shapes.items():
384
+ total_params += 1
385
+ unpartitioned_numel = shape.numel()
386
+ total_numel += unpartitioned_numel
387
+
388
+ param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
389
+ state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
390
+
391
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
392
+
393
+ if debug:
394
+ print(
395
+ f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
396
+ )
397
+
398
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
399
+
400
+
401
+ def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
402
+ param_shapes = zero_model_states[0].param_shapes
403
+ avail_numel = fp32_flat_groups[0].numel() * world_size
404
+ # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
405
+ # param, re-consolidating each param, while dealing with padding if any
406
+
407
+ # merge list of dicts, preserving order
408
+ param_shapes = {k: v for d in param_shapes for k, v in d.items()}
409
+
410
+ if debug:
411
+ for i in range(world_size):
412
+ print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
413
+
414
+ wanted_params = len(param_shapes)
415
+ wanted_numel = sum(shape.numel() for shape in param_shapes.values())
416
+ # not asserting if there is a mismatch due to possible padding
417
+ avail_numel = fp32_flat_groups[0].numel() * world_size
418
+ print(f"Trainable params: Have {avail_numel} numels to process.")
419
+ print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
420
+
421
+ # params
422
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
423
+ # out-of-core computing solution
424
+ offset = 0
425
+ total_numel = 0
426
+ total_params = 0
427
+ for name, shape in tqdm(param_shapes.items(), desc='Gathering Sharded Weights'):
428
+ unpartitioned_numel = shape.numel()
429
+ total_numel += unpartitioned_numel
430
+ total_params += 1
431
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
432
+
433
+ if debug:
434
+ print(
435
+ f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
436
+ )
437
+
438
+ # XXX: memory usage doubles here
439
+ state_dict[name] = torch.cat(
440
+ tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
441
+ 0).narrow(0, 0, unpartitioned_numel).view(shape)
442
+ offset += partitioned_numel
443
+
444
+ offset *= world_size
445
+
446
+ # Sanity check
447
+ if offset != avail_numel:
448
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
449
+
450
+ print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
451
+
452
+
453
+ def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
454
+ exclude_frozen_parameters):
455
+ state_dict = OrderedDict()
456
+
457
+ # buffers
458
+ buffers = zero_model_states[0].buffers
459
+ state_dict.update(buffers)
460
+ if debug:
461
+ print(f"added {len(buffers)} buffers")
462
+
463
+ if not exclude_frozen_parameters:
464
+ _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
465
+
466
+ _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
467
+
468
+ # recover shared parameters
469
+ for pair in zero_model_states[0].shared_params:
470
+ if pair[1] in state_dict:
471
+ state_dict[pair[0]] = state_dict[pair[1]]
472
+
473
+ return state_dict
474
+
475
+
476
+ def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False):
477
+ """
478
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
479
+ ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
480
+ via a model hub.
481
+
482
+ Args:
483
+ - ``checkpoint_dir``: path to the desired checkpoint folder
484
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
485
+ - ``exclude_frozen_parameters``: exclude frozen parameters
486
+
487
+ Returns:
488
+ - pytorch ``state_dict``
489
+
490
+ Note: this approach may not work if your application doesn't have sufficient free CPU memory and
491
+ you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
492
+ the checkpoint.
493
+
494
+ A typical usage might be ::
495
+
496
+ from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
497
+ # do the training and checkpoint saving
498
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
499
+ model = model.cpu() # move to cpu
500
+ model.load_state_dict(state_dict)
501
+ # submit to model hub or save the model to share with others
502
+
503
+ In this example the ``model`` will no longer be usable in the deepspeed context of the same
504
+ application. i.e. you will need to re-initialize the deepspeed engine, since
505
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
506
+
507
+ If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
508
+
509
+ """
510
+ if tag is None:
511
+ latest_path = os.path.join(checkpoint_dir, 'latest')
512
+ if os.path.isfile(latest_path):
513
+ with open(latest_path, 'r') as fd:
514
+ tag = fd.read().strip()
515
+ else:
516
+ raise ValueError(f"Unable to find 'latest' file at {latest_path}")
517
+
518
+ ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
519
+
520
+ if not os.path.isdir(ds_checkpoint_dir):
521
+ raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
522
+
523
+ return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
524
+
525
+
526
+ def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir,
527
+ output_dir,
528
+ max_shard_size="5GB",
529
+ safe_serialization=False,
530
+ tag=None,
531
+ exclude_frozen_parameters=False):
532
+ """
533
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
534
+ loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
535
+
536
+ Args:
537
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
538
+ - ``output_dir``: directory to the pytorch fp32 state_dict output files
539
+ - ``max_shard_size``: the maximum size for a checkpoint before being sharded, default value is 5GB
540
+ - ``safe_serialization``: whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).
541
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
542
+ - ``exclude_frozen_parameters``: exclude frozen parameters
543
+ """
544
+ # Dependency pre-check
545
+ if safe_serialization:
546
+ try:
547
+ from safetensors.torch import save_file
548
+ except ImportError:
549
+ print('If you want to use `safe_serialization`, please `pip install safetensors`')
550
+ raise
551
+ if max_shard_size is not None:
552
+ try:
553
+ from huggingface_hub import split_torch_state_dict_into_shards
554
+ except ImportError:
555
+ print('If you want to use `max_shard_size`, please `pip install huggingface_hub`')
556
+ raise
557
+
558
+ # Convert zero checkpoint to state_dict
559
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters)
560
+
561
+ # Shard the model if it is too big.
562
+ weights_name = "model.safetensors" if safe_serialization else "pytorch_model.bin"
563
+ if max_shard_size is not None:
564
+ filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(".safetensors", "{suffix}.safetensors")
565
+ state_dict_split = split_torch_state_dict_into_shards(state_dict,
566
+ filename_pattern=filename_pattern,
567
+ max_shard_size=max_shard_size)
568
+ else:
569
+ from collections import namedtuple
570
+ StateDictSplit = namedtuple("StateDictSplit", ["is_sharded", "filename_to_tensors"])
571
+ state_dict_split = StateDictSplit(is_sharded=False,
572
+ filename_to_tensors={weights_name: list(state_dict.keys())})
573
+
574
+ # Save the model
575
+ filename_to_tensors = state_dict_split.filename_to_tensors.items()
576
+ for shard_file, tensors in tqdm(filename_to_tensors, desc="Saving checkpoint shards"):
577
+ shard = {tensor: state_dict[tensor].contiguous() for tensor in tensors}
578
+ output_path = os.path.join(output_dir, shard_file)
579
+ if safe_serialization:
580
+ save_file(shard, output_path, metadata={"format": "pt"})
581
+ else:
582
+ torch.save(shard, output_path)
583
+
584
+ # Save index if sharded
585
+ if state_dict_split.is_sharded:
586
+ index = {
587
+ "metadata": state_dict_split.metadata,
588
+ "weight_map": state_dict_split.tensor_to_filename,
589
+ }
590
+ save_index_file = "model.safetensors.index.json" if safe_serialization else "pytorch_model.bin.index.json"
591
+ save_index_file = os.path.join(output_dir, save_index_file)
592
+ with open(save_index_file, "w", encoding="utf-8") as f:
593
+ content = json.dumps(index, indent=2, sort_keys=True) + "\n"
594
+ f.write(content)
595
+
596
+
597
+ def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
598
+ """
599
+ 1. Put the provided model to cpu
600
+ 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
601
+ 3. Load it into the provided model
602
+
603
+ Args:
604
+ - ``model``: the model object to update
605
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
606
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
607
+
608
+ Returns:
609
+ - ``model`: modified model
610
+
611
+ Make sure you have plenty of CPU memory available before you call this function. If you don't
612
+ have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
613
+ conveniently placed for you in the checkpoint folder.
614
+
615
+ A typical usage might be ::
616
+
617
+ from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
618
+ model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
619
+ # submit to model hub or save the model to share with others
620
+
621
+ Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
622
+ of the same application. i.e. you will need to re-initialize the deepspeed engine, since
623
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
624
+
625
+ """
626
+ logger.info(f"Extracting fp32 weights")
627
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
628
+
629
+ logger.info(f"Overwriting model with fp32 weights")
630
+ model = model.cpu()
631
+ model.load_state_dict(state_dict, strict=False)
632
+
633
+ return model
634
+
635
+
636
+ if __name__ == "__main__":
637
+ parser = argparse.ArgumentParser()
638
+ parser.add_argument("checkpoint_dir",
639
+ type=str,
640
+ help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
641
+ parser.add_argument("output_dir",
642
+ type=str,
643
+ help="directory to the pytorch fp32 state_dict output files"
644
+ "(e.g. path/checkpoint-12-output/)")
645
+ parser.add_argument(
646
+ "--max_shard_size",
647
+ type=str,
648
+ default="5GB",
649
+ help="The maximum size for a checkpoint before being sharded. Checkpoints shard will then be each of size"
650
+ "lower than this size. If expressed as a string, needs to be digits followed by a unit (like `5MB`"
651
+ "We default it to 5GB in order for models to be able to run easily on free-tier google colab instances"
652
+ "without CPU OOM issues.")
653
+ parser.add_argument(
654
+ "--safe_serialization",
655
+ default=False,
656
+ action='store_true',
657
+ help="Whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).")
658
+ parser.add_argument("-t",
659
+ "--tag",
660
+ type=str,
661
+ default=None,
662
+ help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
663
+ parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters")
664
+ parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
665
+ args = parser.parse_args()
666
+
667
+ debug = args.debug
668
+
669
+ convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir,
670
+ args.output_dir,
671
+ max_shard_size=args.max_shard_size,
672
+ safe_serialization=args.safe_serialization,
673
+ tag=args.tag,
674
+ exclude_frozen_parameters=args.exclude_frozen_parameters)