Hugging Face's logo

Pre2Post-Chess-RL
/
Chess-RL-Models

Reinforcement Learning
Safetensors
chess
grpo
Model card Files
xet
 Community
Evangelinejy commited on
Commit
f75c74f
·
verified ·
1 Parent(s): ca6f292

cards: drop alpha/beta/compute/run-id

Browse files
Files changed (2) hide show
  1. README.md +43 -44
  2. model_680m_16B/README.md +0 -4
README.md CHANGED
@@ -45,51 +45,50 @@ tok = AutoTokenizer.from_pretrained(p, trust_remote_code=True)
45
 
46
  ## Models
47
 
48
- | model | size | pretrain tokens | compute | alpha | beta | ckpts | steps | run id |
49
- |---|---|---|---|---|---|---|---|---|
50
- | `model_200m_0.27B` | 200m | 0.267B | 6p5e18 | 0.05 | 0.1 | 10 | 100–1000 | `C6p5e18_200m_alpha0.050_beta0.100` |
51
- | `model_200m_1.1B` | 200m | 1.07B | 6p5e18 | 0.2 | 0.1 | 20 | 100–2000 | `C6p5e18_200m_alpha0.200_beta0.100` |
52
- | `model_200m_2.1B` | 200m | 2.13B | 6p5e18 | 0.4 | 0.1 | 20 | 100–2000 | `C6p5e18_200m_alpha0.400_beta0.100` |
53
- | `model_200m_4.0B` | 200m | 4B | 6p5e18 | 0.75 | 0.1 | 10 | 100–1000 | `C6p5e18_200m_alpha0.750_beta0.100` |
54
- | `model_200m_5.3B` | 200m | 5.33B | 6p5e18 | 1.0 | 0.1 | 20 | 100–2000 | `C6p5e18_200m_alpha1.000_beta0.100` |
55
- | `model_20m_0.53B` | 20m | 0.527B | 6p5e18 | 0.01 | 0.008 | 20 | 100–2000 | `C6p5e18_20m_alpha0.010_beta0.008` |
56
- | `model_20m_1.6B` | 20m | 1.58B | 6p5e18 | 0.03 | 0.008 | 50 | 100–5000 | `C6p5e18_20m_alpha0.030_beta0.008` |
57
- | `model_20m_2.6B` | 20m | 2.64B | 6p5e18 | 0.05 | 0.008 | 49 | 100–4900 | `C6p5e18_20m_alpha0.050_beta0.008` |
58
- | `model_20m_5.3B` | 20m | 5.27B | 6p5e18 | 0.1 | 0.008 | 50 | 100–5000 | `C6p5e18_20m_alpha0.100_beta0.008` |
59
- | `model_20m_11B` | 20m | 10.5B | 6p5e18 | 0.2 | 0.008 | 50 | 100–5000 | `C6p5e18_20m_alpha0.200_beta0.008` |
60
- | `model_20m_16B` | 20m | 15.8B | 6p5e18 | 0.3 | 0.008 | 30 | 100–3000 | `C6p5e18_20m_alpha0.300_beta0.008` |
61
- | `model_20m_21B` | 20m | 21.1B | 6p5e18 | 0.4 | 0.008 | 50 | 100–5000 | `C6p5e18_20m_alpha0.400_beta0.008` |
62
- | `model_20m_32B` | 20m | 31.6B | 6p5e18 | 0.6 | 0.008 | 50 | 100–5000 | `C6p5e18_20m_alpha0.600_beta0.008` |
63
- | `model_20m_40B` | 20m | 39.6B | 6p5e18 | 0.75 | 0.008 | 49 | 100–4900 | `C6p5e18_20m_alpha0.750_beta0.008` |
64
- | `model_20m_53B` | 20m | 52.7B | 6p5e18 | 1.0 | 0.008 | 32 | 100–5000 | `C6p5e18_20m_alpha1.000_beta0.008` |
65
- | `model_50m_0.23B` | 50m | 0.229B | 6p5e18 | 0.01 | 0.023 | 20 | 100–2000 | `C6p5e18_50m_alpha0.010_beta0.023` |
66
- | `model_50m_0.69B` | 50m | 0.687B | 6p5e18 | 0.03 | 0.023 | 12 | 100–1200 | `C6p5e18_50m_alpha0.030_beta0.023` |
67
- | `model_50m_1.1B` | 50m | 1.15B | 6p5e18 | 0.05 | 0.023 | 10 | 100–1000 | `C6p5e18_50m_alpha0.050_beta0.023` |
68
- | `model_50m_2.3B` | 50m | 2.29B | 6p5e18 | 0.1 | 0.023 | 20 | 100–2000 | `C6p5e18_50m_alpha0.100_beta0.023` |
69
- | `model_50m_4.6B` | 50m | 4.58B | 6p5e18 | 0.2 | 0.023 | 20 | 100–2000 | `C6p5e18_50m_alpha0.200_beta0.023` |
70
- | `model_50m_9.2B` | 50m | 9.16B | 6p5e18 | 0.4 | 0.023 | 20 | 100–2000 | `C6p5e18_50m_alpha0.400_beta0.023` |
71
- | `model_50m_17B` | 50m | 17.2B | 6p5e18 | 0.75 | 0.023 | 20 | 100–2000 | `C6p5e18_50m_alpha0.750_beta0.023` |
72
- | `model_50m_23B` | 50m | 22.9B | 6p5e18 | 1.0 | 0.023 | 10 | 100–1000 | `C6p5e18_50m_alpha1.000_beta0.023` |
73
- | `model_50m_41B` | 50m | 41.2B | 6p5e19 | 0.18 | 0.002 | 18 | 100–2000 | `C6p5e19_50m_alpha0.180_beta0.002` |
74
- | `model_680m_0.32B` | 680m | 0.319B | 6p5e18 | 0.2 | 0.296 | 5 | 100–500 | `C6p5e18_680m_alpha0.200_beta0.296` |
75
- | `model_680m_1.6B` | 680m | 1.6B | 6p5e18 | 1.0 | 0.296 | 1 | 2000–2000 | `C6p5e18_680m_alpha1.000_beta0.296` |
76
- | `model_680m_6.4B` | 680m | 6.4B | 6p5e19 | 0.4 | 0.03 | 10 | 100–1000 | `C6p5e19_680m_alpha0.400_beta0.030` |
77
- | `model_680m_16B` | 680m | 16B | 6p5e19 | 1.0 | 0.03 | 4 | 1200–1900 | `C6p5e19_680m_alpha1.000_beta0.030` |
78
 
79
  ## Not included
80
 
81
- These runs exist in the study but had no usable HF-format weights on disk at
82
- upload time (evals survive, the weights were cleaned up):
83
 
84
- - `C6p5e18_200m_alpha0.100_beta0.100` — no usable checkpoint on disk
85
- - `C6p5e18_32m_alpha0.200_beta0.013` — no usable checkpoint on disk
86
- - `C6p5e18_32m_alpha0.400_beta0.013` — not in run_metadata.csv - cannot derive name
87
- - `C6p5e18_680m_alpha0.400_beta0.296` — no usable checkpoint on disk
88
- - `C6p5e18_680m_alpha0.750_beta0.296` — no usable checkpoint on disk
89
- - `C6p5e19_200m_alpha0.200_beta0.007` — no usable checkpoint on disk
90
- - `C6p5e19_200m_alpha0.400_beta0.007` — no usable checkpoint on disk
91
- - `C6p5e19_200m_alpha0.750_beta0.007` — no usable checkpoint on disk
92
- - `C6p5e19_200m_alpha1.000_beta0.007` — no usable checkpoint on disk
93
- - `C6p5e19_680m_alpha0.200_beta0.030` — no usable checkpoint on disk
94
- - `C6p5e19_680m_alpha0.750_beta0.030` — no usable checkpoint on disk
95
- - `C6p5e19_680m_alpha2.000_beta0.030` — no usable checkpoint on disk
 
45
 
46
  ## Models
47
 
48
+ | model | size | pretrain tokens | ckpts | steps |
49
+ |---|---|---|---|---|
50
+ | `model_200m_0.27B` | 200m | 0.267B | 10 | 100–1000 |
51
+ | `model_200m_1.1B` | 200m | 1.07B | 20 | 100–2000 |
52
+ | `model_200m_2.1B` | 200m | 2.13B | 20 | 100–2000 |
53
+ | `model_200m_4.0B` | 200m | 4B | 10 | 100–1000 |
54
+ | `model_200m_5.3B` | 200m | 5.33B | 20 | 100–2000 |
55
+ | `model_20m_0.53B` | 20m | 0.527B | 20 | 100–2000 |
56
+ | `model_20m_1.6B` | 20m | 1.58B | 50 | 100–5000 |
57
+ | `model_20m_2.6B` | 20m | 2.64B | 49 | 100–4900 |
58
+ | `model_20m_5.3B` | 20m | 5.27B | 50 | 100–5000 |
59
+ | `model_20m_11B` | 20m | 10.5B | 50 | 100–5000 |
60
+ | `model_20m_16B` | 20m | 15.8B | 30 | 100–3000 |
61
+ | `model_20m_21B` | 20m | 21.1B | 50 | 100–5000 |
62
+ | `model_20m_32B` | 20m | 31.6B | 50 | 100–5000 |
63
+ | `model_20m_40B` | 20m | 39.6B | 49 | 100–4900 |
64
+ | `model_20m_53B` | 20m | 52.7B | 32 | 100–5000 |
65
+ | `model_50m_0.23B` | 50m | 0.229B | 20 | 100–2000 |
66
+ | `model_50m_0.69B` | 50m | 0.687B | 12 | 100–1200 |
67
+ | `model_50m_1.1B` | 50m | 1.15B | 10 | 100–1000 |
68
+ | `model_50m_2.3B` | 50m | 2.29B | 20 | 100–2000 |
69
+ | `model_50m_4.6B` | 50m | 4.58B | 20 | 100–2000 |
70
+ | `model_50m_9.2B` | 50m | 9.16B | 20 | 100–2000 |
71
+ | `model_50m_17B` | 50m | 17.2B | 20 | 100–2000 |
72
+ | `model_50m_23B` | 50m | 22.9B | 10 | 100–1000 |
73
+ | `model_50m_41B` | 50m | 41.2B | 18 | 100–2000 |
74
+ | `model_680m_0.32B` | 680m | 0.319B | 5 | 100–500 |
75
+ | `model_680m_1.6B` | 680m | 1.6B | 1 | 2000 |
76
+ | `model_680m_6.4B` | 680m | 6.4B | 10 | 100–1000 |
77
+ | `model_680m_16B` | 680m | 16B | 4 | 1200–1900 |
78
 
79
  ## Not included
80
 
81
+ These models exist in the study but had no usable weights available at upload
82
+ time:
83
 
84
+ - `model_200m_0.53B`
85
+ - `model_200m_11B`
86
+ - `model_200m_21B`
87
+ - `model_200m_40B`
88
+ - `model_200m_53B`
89
+ - `model_680m_0.64B`
90
+ - `model_680m_1.2B`
91
+ - `model_680m_12B`
92
+ - `model_680m_3.2B`
93
+ - `model_680m_32B`
94
+ - 2 32m models absent from the pretraining-token tables
 
model_680m_16B/README.md CHANGED
@@ -12,12 +12,8 @@ The pretraining base and SFT init for this model are
12
 
13
  | | |
14
  |---|---|
15
- | run id | `C6p5e19_680m_alpha1.000_beta0.030` |
16
  | parameters | 680m |
17
  | pretraining tokens | 15,994,915,000 (16B) |
18
- | compute class | 6p5e19 |
19
- | alpha (pretrain fraction) | 1.0 |
20
- | beta (SFT fraction) | 0.03 |
21
  | checkpoints here | 4 (steps 1200–1900) |
22
  | checkpoints saved by the run | 25 |
23
 
 
12
 
13
  | | |
14
  |---|---|
 
15
  | parameters | 680m |
16
  | pretraining tokens | 15,994,915,000 (16B) |
 
 
 
17
  | checkpoints here | 4 (steps 1200–1900) |
18
  | checkpoints saved by the run | 25 |
19