kimgreg commited on
Commit
be5d658
·
verified ·
1 Parent(s): 0df0f0a

checkpoints of frz6-fmt8

Browse files
frz6-fmt8/e0.25/config.json ADDED
@@ -0,0 +1,65 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Lfm2ForCausalLM"
4
+ ],
5
+ "block__name_mlp": "parallel_mlp_merged",
6
+ "block_auto_adjust_ff_dim": false,
7
+ "block_dim": 1024,
8
+ "block_ffn_dim_multiplier": 1.0,
9
+ "block_ffn_te_autocast": false,
10
+ "block_ffn_use_quantized_params": false,
11
+ "block_mlp_init_scale": 1.0,
12
+ "block_multiple_of": 256,
13
+ "block_norm_eps": 1e-05,
14
+ "block_out_init_scale": 1.0,
15
+ "block_sequence_parallel_norm_across_tp": false,
16
+ "block_use_swiglu": true,
17
+ "block_use_xavier_init": true,
18
+ "bos_token_id": 1,
19
+ "conv_L_cache": 3,
20
+ "conv_bias": false,
21
+ "conv_dim": 1024,
22
+ "conv_use_xavier_init": true,
23
+ "dtype": "float32",
24
+ "eos_token_id": 7,
25
+ "ffn_te_autocast": false,
26
+ "ffn_use_quantized_params": false,
27
+ "full_attn_idxs": null,
28
+ "hidden_size": 1024,
29
+ "initializer_range": 0.02,
30
+ "intermediate_size": 2560,
31
+ "layer_types": [
32
+ "conv",
33
+ "conv",
34
+ "full_attention",
35
+ "conv",
36
+ "full_attention",
37
+ "conv",
38
+ "full_attention",
39
+ "conv",
40
+ "full_attention",
41
+ "conv",
42
+ "full_attention",
43
+ "conv",
44
+ "full_attention",
45
+ "conv"
46
+ ],
47
+ "max_position_embeddings": 128000,
48
+ "model_type": "lfm2",
49
+ "norm_eps": 1e-05,
50
+ "num_attention_heads": 16,
51
+ "num_heads": 16,
52
+ "num_hidden_layers": 14,
53
+ "num_key_value_heads": 8,
54
+ "pad_token_id": 0,
55
+ "rope_parameters": {
56
+ "rope_theta": 1000000.0,
57
+ "rope_type": "default"
58
+ },
59
+ "sequence_parallel_norm_across_tp": false,
60
+ "tie_word_embeddings": true,
61
+ "transformers_version": "5.22.2",
62
+ "use_cache": true,
63
+ "use_pos_enc": true,
64
+ "vocab_size": 65536
65
+ }
frz6-fmt8/e0.25/generation_config.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "do_sample": true,
5
+ "eos_token_id": 7,
6
+ "output_attentions": false,
7
+ "output_hidden_states": false,
8
+ "pad_token_id": 0,
9
+ "repetition_penalty": 1.05,
10
+ "temperature": 0.1,
11
+ "top_k": 50,
12
+ "transformers_version": "5.22.2",
13
+ "use_cache": true
14
+ }
frz6-fmt8/e0.25/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:424630d3b804ee4164b56d5ccf13e524dbc0aba8a345b22a885c9efd6f12e436
3
+ size 593618936
frz6-fmt8/e1/config.json ADDED
@@ -0,0 +1,65 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Lfm2ForCausalLM"
4
+ ],
5
+ "block__name_mlp": "parallel_mlp_merged",
6
+ "block_auto_adjust_ff_dim": false,
7
+ "block_dim": 1024,
8
+ "block_ffn_dim_multiplier": 1.0,
9
+ "block_ffn_te_autocast": false,
10
+ "block_ffn_use_quantized_params": false,
11
+ "block_mlp_init_scale": 1.0,
12
+ "block_multiple_of": 256,
13
+ "block_norm_eps": 1e-05,
14
+ "block_out_init_scale": 1.0,
15
+ "block_sequence_parallel_norm_across_tp": false,
16
+ "block_use_swiglu": true,
17
+ "block_use_xavier_init": true,
18
+ "bos_token_id": 1,
19
+ "conv_L_cache": 3,
20
+ "conv_bias": false,
21
+ "conv_dim": 1024,
22
+ "conv_use_xavier_init": true,
23
+ "dtype": "float32",
24
+ "eos_token_id": 7,
25
+ "ffn_te_autocast": false,
26
+ "ffn_use_quantized_params": false,
27
+ "full_attn_idxs": null,
28
+ "hidden_size": 1024,
29
+ "initializer_range": 0.02,
30
+ "intermediate_size": 2560,
31
+ "layer_types": [
32
+ "conv",
33
+ "conv",
34
+ "full_attention",
35
+ "conv",
36
+ "full_attention",
37
+ "conv",
38
+ "full_attention",
39
+ "conv",
40
+ "full_attention",
41
+ "conv",
42
+ "full_attention",
43
+ "conv",
44
+ "full_attention",
45
+ "conv"
46
+ ],
47
+ "max_position_embeddings": 128000,
48
+ "model_type": "lfm2",
49
+ "norm_eps": 1e-05,
50
+ "num_attention_heads": 16,
51
+ "num_heads": 16,
52
+ "num_hidden_layers": 14,
53
+ "num_key_value_heads": 8,
54
+ "pad_token_id": 0,
55
+ "rope_parameters": {
56
+ "rope_theta": 1000000.0,
57
+ "rope_type": "default"
58
+ },
59
+ "sequence_parallel_norm_across_tp": false,
60
+ "tie_word_embeddings": true,
61
+ "transformers_version": "5.22.2",
62
+ "use_cache": true,
63
+ "use_pos_enc": true,
64
+ "vocab_size": 65536
65
+ }
frz6-fmt8/e1/generation_config.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "do_sample": true,
5
+ "eos_token_id": 7,
6
+ "output_attentions": false,
7
+ "output_hidden_states": false,
8
+ "pad_token_id": 0,
9
+ "repetition_penalty": 1.05,
10
+ "temperature": 0.1,
11
+ "top_k": 50,
12
+ "transformers_version": "5.22.2",
13
+ "use_cache": true
14
+ }
frz6-fmt8/e1/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6f37b1aaf2617669f5d72e6e7ecb797903de4ae9064ce462abb61a333a620c0c
3
+ size 593618936
frz6-fmt8/e20/config.json ADDED
@@ -0,0 +1,65 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Lfm2ForCausalLM"
4
+ ],
5
+ "block__name_mlp": "parallel_mlp_merged",
6
+ "block_auto_adjust_ff_dim": false,
7
+ "block_dim": 1024,
8
+ "block_ffn_dim_multiplier": 1.0,
9
+ "block_ffn_te_autocast": false,
10
+ "block_ffn_use_quantized_params": false,
11
+ "block_mlp_init_scale": 1.0,
12
+ "block_multiple_of": 256,
13
+ "block_norm_eps": 1e-05,
14
+ "block_out_init_scale": 1.0,
15
+ "block_sequence_parallel_norm_across_tp": false,
16
+ "block_use_swiglu": true,
17
+ "block_use_xavier_init": true,
18
+ "bos_token_id": 1,
19
+ "conv_L_cache": 3,
20
+ "conv_bias": false,
21
+ "conv_dim": 1024,
22
+ "conv_use_xavier_init": true,
23
+ "dtype": "float32",
24
+ "eos_token_id": 7,
25
+ "ffn_te_autocast": false,
26
+ "ffn_use_quantized_params": false,
27
+ "full_attn_idxs": null,
28
+ "hidden_size": 1024,
29
+ "initializer_range": 0.02,
30
+ "intermediate_size": 2560,
31
+ "layer_types": [
32
+ "conv",
33
+ "conv",
34
+ "full_attention",
35
+ "conv",
36
+ "full_attention",
37
+ "conv",
38
+ "full_attention",
39
+ "conv",
40
+ "full_attention",
41
+ "conv",
42
+ "full_attention",
43
+ "conv",
44
+ "full_attention",
45
+ "conv"
46
+ ],
47
+ "max_position_embeddings": 128000,
48
+ "model_type": "lfm2",
49
+ "norm_eps": 1e-05,
50
+ "num_attention_heads": 16,
51
+ "num_heads": 16,
52
+ "num_hidden_layers": 14,
53
+ "num_key_value_heads": 8,
54
+ "pad_token_id": 0,
55
+ "rope_parameters": {
56
+ "rope_theta": 1000000.0,
57
+ "rope_type": "default"
58
+ },
59
+ "sequence_parallel_norm_across_tp": false,
60
+ "tie_word_embeddings": true,
61
+ "transformers_version": "5.22.2",
62
+ "use_cache": true,
63
+ "use_pos_enc": true,
64
+ "vocab_size": 65536
65
+ }
frz6-fmt8/e20/generation_config.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "do_sample": true,
5
+ "eos_token_id": 7,
6
+ "output_attentions": false,
7
+ "output_hidden_states": false,
8
+ "pad_token_id": 0,
9
+ "repetition_penalty": 1.05,
10
+ "temperature": 0.1,
11
+ "top_k": 50,
12
+ "transformers_version": "5.22.2",
13
+ "use_cache": true
14
+ }
frz6-fmt8/e20/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9b22a7345b0440fba41623743b166011392e9ea0732187311b50856fd590cb3d
3
+ size 593618936
frz6-fmt8/e3/config.json ADDED
@@ -0,0 +1,65 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Lfm2ForCausalLM"
4
+ ],
5
+ "block__name_mlp": "parallel_mlp_merged",
6
+ "block_auto_adjust_ff_dim": false,
7
+ "block_dim": 1024,
8
+ "block_ffn_dim_multiplier": 1.0,
9
+ "block_ffn_te_autocast": false,
10
+ "block_ffn_use_quantized_params": false,
11
+ "block_mlp_init_scale": 1.0,
12
+ "block_multiple_of": 256,
13
+ "block_norm_eps": 1e-05,
14
+ "block_out_init_scale": 1.0,
15
+ "block_sequence_parallel_norm_across_tp": false,
16
+ "block_use_swiglu": true,
17
+ "block_use_xavier_init": true,
18
+ "bos_token_id": 1,
19
+ "conv_L_cache": 3,
20
+ "conv_bias": false,
21
+ "conv_dim": 1024,
22
+ "conv_use_xavier_init": true,
23
+ "dtype": "float32",
24
+ "eos_token_id": 7,
25
+ "ffn_te_autocast": false,
26
+ "ffn_use_quantized_params": false,
27
+ "full_attn_idxs": null,
28
+ "hidden_size": 1024,
29
+ "initializer_range": 0.02,
30
+ "intermediate_size": 2560,
31
+ "layer_types": [
32
+ "conv",
33
+ "conv",
34
+ "full_attention",
35
+ "conv",
36
+ "full_attention",
37
+ "conv",
38
+ "full_attention",
39
+ "conv",
40
+ "full_attention",
41
+ "conv",
42
+ "full_attention",
43
+ "conv",
44
+ "full_attention",
45
+ "conv"
46
+ ],
47
+ "max_position_embeddings": 128000,
48
+ "model_type": "lfm2",
49
+ "norm_eps": 1e-05,
50
+ "num_attention_heads": 16,
51
+ "num_heads": 16,
52
+ "num_hidden_layers": 14,
53
+ "num_key_value_heads": 8,
54
+ "pad_token_id": 0,
55
+ "rope_parameters": {
56
+ "rope_theta": 1000000.0,
57
+ "rope_type": "default"
58
+ },
59
+ "sequence_parallel_norm_across_tp": false,
60
+ "tie_word_embeddings": true,
61
+ "transformers_version": "5.22.2",
62
+ "use_cache": true,
63
+ "use_pos_enc": true,
64
+ "vocab_size": 65536
65
+ }
frz6-fmt8/e3/generation_config.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "do_sample": true,
5
+ "eos_token_id": 7,
6
+ "output_attentions": false,
7
+ "output_hidden_states": false,
8
+ "pad_token_id": 0,
9
+ "repetition_penalty": 1.05,
10
+ "temperature": 0.1,
11
+ "top_k": 50,
12
+ "transformers_version": "5.22.2",
13
+ "use_cache": true
14
+ }
frz6-fmt8/e3/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a9d1e191f6b56c27d6543f6d20bfa63b8659d6e3dfbe2a4dd6e3d606edce3cd5
3
+ size 593618936
frz6-fmt8/e8/config.json ADDED
@@ -0,0 +1,65 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Lfm2ForCausalLM"
4
+ ],
5
+ "block__name_mlp": "parallel_mlp_merged",
6
+ "block_auto_adjust_ff_dim": false,
7
+ "block_dim": 1024,
8
+ "block_ffn_dim_multiplier": 1.0,
9
+ "block_ffn_te_autocast": false,
10
+ "block_ffn_use_quantized_params": false,
11
+ "block_mlp_init_scale": 1.0,
12
+ "block_multiple_of": 256,
13
+ "block_norm_eps": 1e-05,
14
+ "block_out_init_scale": 1.0,
15
+ "block_sequence_parallel_norm_across_tp": false,
16
+ "block_use_swiglu": true,
17
+ "block_use_xavier_init": true,
18
+ "bos_token_id": 1,
19
+ "conv_L_cache": 3,
20
+ "conv_bias": false,
21
+ "conv_dim": 1024,
22
+ "conv_use_xavier_init": true,
23
+ "dtype": "float32",
24
+ "eos_token_id": 7,
25
+ "ffn_te_autocast": false,
26
+ "ffn_use_quantized_params": false,
27
+ "full_attn_idxs": null,
28
+ "hidden_size": 1024,
29
+ "initializer_range": 0.02,
30
+ "intermediate_size": 2560,
31
+ "layer_types": [
32
+ "conv",
33
+ "conv",
34
+ "full_attention",
35
+ "conv",
36
+ "full_attention",
37
+ "conv",
38
+ "full_attention",
39
+ "conv",
40
+ "full_attention",
41
+ "conv",
42
+ "full_attention",
43
+ "conv",
44
+ "full_attention",
45
+ "conv"
46
+ ],
47
+ "max_position_embeddings": 128000,
48
+ "model_type": "lfm2",
49
+ "norm_eps": 1e-05,
50
+ "num_attention_heads": 16,
51
+ "num_heads": 16,
52
+ "num_hidden_layers": 14,
53
+ "num_key_value_heads": 8,
54
+ "pad_token_id": 0,
55
+ "rope_parameters": {
56
+ "rope_theta": 1000000.0,
57
+ "rope_type": "default"
58
+ },
59
+ "sequence_parallel_norm_across_tp": false,
60
+ "tie_word_embeddings": true,
61
+ "transformers_version": "5.22.2",
62
+ "use_cache": true,
63
+ "use_pos_enc": true,
64
+ "vocab_size": 65536
65
+ }
frz6-fmt8/e8/generation_config.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "do_sample": true,
5
+ "eos_token_id": 7,
6
+ "output_attentions": false,
7
+ "output_hidden_states": false,
8
+ "pad_token_id": 0,
9
+ "repetition_penalty": 1.05,
10
+ "temperature": 0.1,
11
+ "top_k": 50,
12
+ "transformers_version": "5.22.2",
13
+ "use_cache": true
14
+ }
frz6-fmt8/e8/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b80ba96c53516e0ee0daf817be910262ff64b4bb03660d1b9c225ab4b6da0bcc
3
+ size 593618936
frz6-fmt8/run-config.json ADDED
@@ -0,0 +1,45 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model": ".local/models/lfm2.5-230m",
3
+ "kb": "data/kb/kb-v1",
4
+ "data": "data/synth/kb-v1-fmt-k8-short",
5
+ "out": "results/runs/frz6-fmt8",
6
+ "epochs": 20,
7
+ "save_at": "0.25,1,3,8,20",
8
+ "objective": "sft",
9
+ "objective_args": "{}",
10
+ "cpt": false,
11
+ "mixed": false,
12
+ "replay": null,
13
+ "replay_ratio": 0.0,
14
+ "replay_per_epoch": 0,
15
+ "replay_seed": null,
16
+ "stream": 0,
17
+ "lr": 1e-05,
18
+ "beta2": 0.999,
19
+ "weight_decay": 0.0,
20
+ "warmup": 0.03,
21
+ "clip": 1.0,
22
+ "batch": 16,
23
+ "seed": 1,
24
+ "freeze_embeddings": false,
25
+ "freeze_layers": 6,
26
+ "lora_rank": 0,
27
+ "lora_alpha": 16.0,
28
+ "lora_dropout": 0.1,
29
+ "lora_targets": "q_proj,v_proj",
30
+ "start_model": null,
31
+ "token_budget": 4096,
32
+ "trace": false,
33
+ "trace_every": 0,
34
+ "save_state_at": "",
35
+ "resume": null,
36
+ "branch_steps": 0,
37
+ "save_every_steps": 0,
38
+ "reset_optimizer": false,
39
+ "order": "mixed",
40
+ "order_block": 256,
41
+ "max_steps": 0,
42
+ "data_manifest": {
43
+ "train/pairs.jsonl": "8bc0feb8fe0e2f8630a1e4615fcf6cd50239bd66c75b48dc74817bb24c8b8a49"
44
+ }
45
+ }