Hugging Face's logo

enosislabs
/
AETHER-Mythos-1-1.2B

Text Generation
Transformers
Safetensors
English
lfm2
lfm
liquidai
lfm2.5
coding-agent
thinking
fable
aether-mythos
agentic
tool-use
unsloth
conversational
Model card Files
xet
 Community
ferxalb commited on
Commit
6986094
·
verified ·
1 Parent(s): b09ca7b

Upload AETHER-Mythos-1 merged 16-bit weights

Browse files
Files changed (5) hide show
  1. README.md +9 -9
  2. config.json +60 -57
  3. generation_config.json +8 -0
  4. model.safetensors +1 -1
  5. tokenizer_config.json +0 -0
README.md CHANGED
@@ -43,15 +43,15 @@ into an efficient backbone. We prioritize:
43
  ## Base model
44
 
45
  - **Base**: [`LiquidAI/LFM2.5-1.2B-Thinking`](https://huggingface.co/LiquidAI/LFM2.5-1.2B-Thinking)
46
- - **Context trained**: up to 8192 tokens (packing enabled)
47
- - **Method**: LoRA (r=64, alpha=128)
48
 
49
  ## Data mixture
50
 
51
- - **fable5_cot** (`Glint-Research/Fable-5-traces`) weight=0.65 — Primary high-signal Fable 5 CoT + tools (AGPL-3.0)
52
- - **opencode_reasoning** (`nvidia/OpenCodeReasoning`) weight=0.18 — Coding CoT already uses <think> — aligns with LFM Thinking
53
- - **open_r1_codeforces** (`open-r1/codeforces-cots`) weight=0.10 — messages
54
- - **openthoughts_code** (`open-thoughts/OpenThoughts-114k`) weight=0.07 — messages
55
 
56
  ### Data provenance & licenses
57
 
@@ -72,10 +72,10 @@ under Liquid AI’s LFM license terms.
72
  | Setting | Value |
73
  |--------|--------|
74
  | GPU | L40S (Modal) |
75
- | Effective batch size | 32 |
76
- | Learning rate | 0.0001 |
77
  | Schedule | cosine |
78
- | Epochs / max steps | 1.0 / 600 |
79
  | Packing | True |
80
  | Optim | adamw_8bit |
81
  | Grad checkpointing | unsloth |
 
43
  ## Base model
44
 
45
  - **Base**: [`LiquidAI/LFM2.5-1.2B-Thinking`](https://huggingface.co/LiquidAI/LFM2.5-1.2B-Thinking)
46
+ - **Context trained**: up to 32768 tokens (packing enabled)
47
+ - **Method**: LoRA (r=128, alpha=256)
48
 
49
  ## Data mixture
50
 
51
+ - **fable5_cot** (`Glint-Research/Fable-5-traces`) weight=0.65 — Primary identity and agent trace signal
52
+ - **opencode_reasoning** (`nvidia/OpenCodeReasoning`) weight=0.18 — prompt_completion
53
+ - **open_r1_codeforces** (`open-r1/codeforces-cots`) weight=0.07 — messages
54
+ - **openthoughts_code** (`open-thoughts/OpenThoughts-114k`) weight=0.10 — messages
55
 
56
  ### Data provenance & licenses
57
 
 
72
  | Setting | Value |
73
  |--------|--------|
74
  | GPU | L40S (Modal) |
75
+ | Effective batch size | 16 |
76
+ | Learning rate | 8e-05 |
77
  | Schedule | cosine |
78
+ | Epochs / max steps | 1.0 / 100 |
79
  | Packing | True |
80
  | Optim | adamw_8bit |
81
  | Grad checkpointing | unsloth |
config.json CHANGED
@@ -1,58 +1,61 @@
1
  {
2
- "architectures": [
3
- "Lfm2ForCausalLM"
4
- ],
5
- "block_auto_adjust_ff_dim": true,
6
- "block_dim": 2048,
7
- "block_ff_dim": 12288,
8
- "block_ffn_dim_multiplier": 1.0,
9
- "block_mlp_init_scale": 1.0,
10
- "block_multiple_of": 256,
11
- "block_norm_eps": 1e-05,
12
- "block_out_init_scale": 1.0,
13
- "block_use_swiglu": true,
14
- "block_use_xavier_init": true,
15
- "bos_token_id": 1,
16
- "conv_L_cache": 3,
17
- "conv_bias": false,
18
- "conv_dim": 2048,
19
- "conv_use_xavier_init": true,
20
- "torch_dtype": "bfloat16",
21
- "eos_token_id": 7,
22
- "hidden_size": 2048,
23
- "initializer_range": 0.02,
24
- "intermediate_size": 12288,
25
- "layer_types": [
26
- "conv",
27
- "conv",
28
- "full_attention",
29
- "conv",
30
- "conv",
31
- "full_attention",
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
- ],
43
- "max_position_embeddings": 128000,
44
- "model_type": "lfm2",
45
- "norm_eps": 1e-05,
46
- "num_attention_heads": 32,
47
- "num_heads": 32,
48
- "num_hidden_layers": 16,
49
- "num_key_value_heads": 8,
50
- "pad_token_id": 0,
51
- "rope_theta": 1000000.0,
52
- "tie_embedding": true,
53
- "transformers_version": "4.57.2",
54
- "unsloth_version": "2025.11.1",
55
- "use_cache": true,
56
- "use_pos_enc": true,
57
- "vocab_size": 65536
58
- }
 
 
 
 
1
  {
2
+ "architectures": [
3
+ "Lfm2ForCausalLM"
4
+ ],
5
+ "block_auto_adjust_ff_dim": true,
6
+ "block_dim": 2048,
7
+ "block_ffn_dim_multiplier": 1.0,
8
+ "block_mlp_init_scale": 1.0,
9
+ "block_multiple_of": 256,
10
+ "block_norm_eps": 1e-05,
11
+ "block_out_init_scale": 1.0,
12
+ "block_use_swiglu": true,
13
+ "block_use_xavier_init": true,
14
+ "bos_token_id": 1,
15
+ "conv_L_cache": 3,
16
+ "conv_bias": false,
17
+ "conv_dim": 2048,
18
+ "conv_use_xavier_init": true,
19
+ "torch_dtype": "bfloat16",
20
+ "eos_token_id": 7,
21
+ "full_attn_idxs": null,
22
+ "hidden_size": 2048,
23
+ "initializer_range": 0.02,
24
+ "intermediate_size": 12288,
25
+ "layer_types": [
26
+ "conv",
27
+ "conv",
28
+ "full_attention",
29
+ "conv",
30
+ "conv",
31
+ "full_attention",
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
+ ],
43
+ "max_position_embeddings": 128000,
44
+ "model_name": "LiquidAI/LFM2.5-1.2B-Thinking",
45
+ "model_type": "lfm2",
46
+ "norm_eps": 1e-05,
47
+ "num_attention_heads": 32,
48
+ "num_heads": 32,
49
+ "num_hidden_layers": 16,
50
+ "num_key_value_heads": 8,
51
+ "pad_token_id": 0,
52
+ "rope_parameters": {
53
+ "rope_theta": 1000000.0,
54
+ "rope_type": "default"
55
+ },
56
+ "tie_word_embeddings": true,
57
+ "unsloth_version": "2026.7.5",
58
+ "use_cache": false,
59
+ "use_pos_enc": true,
60
+ "vocab_size": 65536
61
+ }
generation_config.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": 7,
5
+ "max_length": 128000,
6
+ "pad_token_id": 0,
7
+ "transformers_version": "5.5.0"
8
+ }
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:d8758fd670a15ab9e210ddd1dd02714d527a478a9769af26cab151d559d500c3
3
  size 2340697936
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:81658e293cdcfd51b6faf3ac75bc3eefbf371bd03cb69ca2e3d7ea6dbdb03fc4
3
  size 2340697936
tokenizer_config.json CHANGED
The diff for this file is too large to render. See raw diff