Hugging Face's logo

meihant
/
AnonymousModel-mht

Text Generation
Transformers
Safetensors
English
qwen3_5
pretraining
Model card Files
xet
 Community
meihant commited on
Commit
4e6ea75
·
verified ·
1 Parent(s): 5740c76

Add models/FCP-SmolLM2-360M-20B/checkpoints/4B artifacts 0-6

Browse files
Files changed (7) hide show
  1. models/FCP-SmolLM2-360M-20B/checkpoints/4B/chat_template.jinja +8 -0
  2. models/FCP-SmolLM2-360M-20B/checkpoints/4B/config.json +34 -0
  3. models/FCP-SmolLM2-360M-20B/checkpoints/4B/generation_config.json +10 -0
  4. models/FCP-SmolLM2-360M-20B/checkpoints/4B/model.safetensors +3 -0
  5. models/FCP-SmolLM2-360M-20B/checkpoints/4B/tokenizer.json +0 -0
  6. models/FCP-SmolLM2-360M-20B/checkpoints/4B/tokenizer_config.json +15 -0
  7. models/FCP-SmolLM2-360M-20B/checkpoints/4B/training_progress.json +65 -0
models/FCP-SmolLM2-360M-20B/checkpoints/4B/chat_template.jinja ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {% for message in messages %}{% if message['role'] == 'system' %}<|im_start|>system
2
+ {{ message['content'] }}<|im_end|>
3
+ {% elif message['role'] == 'user' %}<|im_start|>user
4
+ {{ message['content'] }}<|im_end|>
5
+ {% elif message['role'] == 'assistant' %}<|im_start|>assistant
6
+ {{ message['content'] }}<|im_end|>
7
+ {% endif %}{% endfor %}{% if add_generation_prompt %}<|im_start|>assistant
8
+ {% endif %}
models/FCP-SmolLM2-360M-20B/checkpoints/4B/config.json ADDED
@@ -0,0 +1,34 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "LlamaForCausalLM"
4
+ ],
5
+ "attention_bias": false,
6
+ "attention_dropout": 0.0,
7
+ "bos_token_id": 0,
8
+ "dtype": "bfloat16",
9
+ "eos_token_id": 0,
10
+ "head_dim": 64,
11
+ "hidden_act": "silu",
12
+ "hidden_size": 960,
13
+ "initializer_range": 0.02,
14
+ "intermediate_size": 2560,
15
+ "is_llama_config": true,
16
+ "max_position_embeddings": 8192,
17
+ "mlp_bias": false,
18
+ "model_type": "llama",
19
+ "num_attention_heads": 15,
20
+ "num_hidden_layers": 32,
21
+ "num_key_value_heads": 5,
22
+ "pad_token_id": 0,
23
+ "pretraining_tp": 1,
24
+ "rms_norm_eps": 1e-05,
25
+ "rope_interleaved": false,
26
+ "rope_parameters": {
27
+ "rope_theta": 100000,
28
+ "rope_type": "default"
29
+ },
30
+ "tie_word_embeddings": true,
31
+ "transformers_version": "5.6.2",
32
+ "use_cache": true,
33
+ "vocab_size": 49158
34
+ }
models/FCP-SmolLM2-360M-20B/checkpoints/4B/generation_config.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 0,
4
+ "eos_token_id": 0,
5
+ "output_attentions": false,
6
+ "output_hidden_states": false,
7
+ "pad_token_id": 0,
8
+ "transformers_version": "5.6.2",
9
+ "use_cache": true
10
+ }
models/FCP-SmolLM2-360M-20B/checkpoints/4B/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b1fdcb552b0cf2a575c715f666f03df85bfe6ead2f7d608da4984de885da46bd
3
+ size 723686432
models/FCP-SmolLM2-360M-20B/checkpoints/4B/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
models/FCP-SmolLM2-360M-20B/checkpoints/4B/tokenizer_config.json ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "backend": "tokenizers",
4
+ "bos_token": "<|endoftext|>",
5
+ "clean_up_tokenization_spaces": false,
6
+ "eos_token": "<|endoftext|>",
7
+ "errors": "replace",
8
+ "is_local": true,
9
+ "local_files_only": true,
10
+ "model_max_length": 8192,
11
+ "pad_token": "<|endoftext|>",
12
+ "tokenizer_class": "GPT2Tokenizer",
13
+ "unk_token": "<|endoftext|>",
14
+ "vocab_size": 49152
15
+ }
models/FCP-SmolLM2-360M-20B/checkpoints/4B/training_progress.json ADDED
@@ -0,0 +1,65 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "passed": true,
3
+ "step": 19041,
4
+ "body_tokens": 4000000355,
5
+ "source_tokens": 4000000388,
6
+ "training_tokens": 9938886344,
7
+ "job_id": "2724341",
8
+ "checkpoint_unix_time": 1793984247.805995,
9
+ "training_seconds": 38792.24541464588,
10
+ "training_gpu_hours": 43.10249490679656,
11
+ "milestone": {
12
+ "segment": 1,
13
+ "end_row": 9802903,
14
+ "source_tokens": 4000000388,
15
+ "target_tokens": 4000000000,
16
+ "start_pack": 1218688,
17
+ "end_pack": 2437128
18
+ },
19
+ "config": {
20
+ "seed": 42,
21
+ "window": 4096,
22
+ "world_size": 4,
23
+ "global_batch": 128,
24
+ "learning_rate": 0.0005,
25
+ "warmup_steps": 2000,
26
+ "scheduler": "constant_with_warmup",
27
+ "adam_betas": [
28
+ 0.9,
29
+ 0.95
30
+ ],
31
+ "adam_epsilon": 1e-08,
32
+ "weight_decay": 0.01,
33
+ "max_grad_norm": 1.0,
34
+ "save_body_tokens": 2000000000,
35
+ "initialization": "seed42 random SmolLM2-360M architecture, shared expanded vocabulary; no pretrained weights",
36
+ "precision": "BF16 parameters, gradients and AdamW moments; fused linear CE",
37
+ "loss": "global supervised-token mean, summed loss normalized before gradient accumulation",
38
+ "packing": "128 global 4096 BFD packs/update, 1000-record candidate buffers; independent source-fragment attention and reset positions; final partial update",
39
+ "data": "ten shared native-token 2B source segments, stop at 20B; source order and BFD buffers stay inside each segment",
40
+ "endings": "Vanilla native EOS; FCP ChatML im_end newline; historical 4096 truncation",
41
+ "checkpoint": "inference-only permanent at each shared source segment end; rolling recovery every 1000 steps (latest 1); 20B endpoint only",
42
+ "repo": "meihant/AnonymousModel-mht"
43
+ },
44
+ "code_hashes": {
45
+ "bfd.py": "ba37e45bfe6882b7c1747bfaafbaf55611f0cdfa2971ad16cd58d88a4009640d",
46
+ "config.py": "6b617d0a45048a8851b617f29608602dc7e290166749dd1551433033c44ec214",
47
+ "encoding.py": "3aed0431713870e35595f922f22be6a4cce123877f493de5bb1f848809683ab4",
48
+ "fcp_format.py": "bef4a540107f949af0e9a2149419ea05cdc4f52385b66eaed84ef25dc40a8305",
49
+ "gate.py": "6c862bfac6e1bd38988cd161804ff9ecb16418e08a031992dfa2f1f78a99ef37",
50
+ "layout.py": "f30a8a77cb97a4bf8d33248a14bbc9112c358b0a0162255d7a77abb15f8f7e2b",
51
+ "model.py": "f0fb8a8789fbf0cd2d7033cf813fa40dc4ebf47617f1f3b70dafa419842eeb5e",
52
+ "prepare_packs.py": "9111f8372105c8381c3b46cd6f31e38ba8d0c4a2c7afde28f4b5a8215c0a813e",
53
+ "runtime.py": "628f7ab759bc4e1047e86dbe71f143fdffa54f16ed498217e7abe82c114c1f7a",
54
+ "train.py": "ea3a7991373ca107118011db93c15e3c18d8f94695909ec26d686ab228e292b3",
55
+ "upload.py": "747a9bce54eafc35b1fa14ba5aa653908bf8f81f24b5e526bce9c947b4ef3182",
56
+ "validate_data.py": "f8dec33d7fc01b54b562ede8cfb87ecbf529dd68be8b48cf9147e6bd3a9324fa",
57
+ "verify.py": "cec4349e3b7d5adcab0f4377b0416cc87b6d810c749833e7b455dbd7e9462a97",
58
+ "checkpoint_files.py": "3a4cd9b1cc858e9c48dc5c00b92a1d4e86330643d03fdb9df361c6ed85788155"
59
+ },
60
+ "final": false,
61
+ "method": "fcp",
62
+ "state_signature": null,
63
+ "checkpoint_kind": "evaluation",
64
+ "resumable": false
65
+ }