Update README.md
Browse files
README.md
CHANGED
|
@@ -1,11 +1,12 @@
|
|
| 1 |
---
|
| 2 |
license: apache-2.0
|
| 3 |
datasets:
|
| 4 |
-
|
| 5 |
library_name: transformers
|
| 6 |
tags:
|
| 7 |
-
|
| 8 |
pipeline_tag: text-generation
|
|
|
|
| 9 |
---
|
| 10 |
|
| 11 |
# ChessSLM
|
|
@@ -36,8 +37,8 @@ ChessSLM demonstrates that **specialized small language models can perform compe
|
|
| 36 |
ChessSLM can play chess by generating moves sequentially in SAN notation.
|
| 37 |
It has been evaluated in matches against several language models, including:
|
| 38 |
|
| 39 |
-
- Claude
|
| 40 |
-
- Gemini
|
| 41 |
- Qwen
|
| 42 |
- GPT-2
|
| 43 |
- GPT-Neo
|
|
@@ -46,7 +47,7 @@ It has been evaluated in matches against several language models, including:
|
|
| 46 |
- Mistral
|
| 47 |
- other small chess-oriented models
|
| 48 |
|
| 49 |
-
The model achieves an
|
| 50 |
|
| 51 |
---
|
| 52 |
|
|
@@ -54,18 +55,18 @@ The model achieves an **Elo rating of approximately 1087**, averaging **around ~
|
|
| 54 |
|
| 55 |
| Model | Elo Rating |
|
| 56 |
|------|------------|
|
| 57 |
-
| EleutherAI/pythia-70m-deduped |
|
|
|
|
| 58 |
| nlpguy/amdchess-v9 | 1094 |
|
| 59 |
| nlpguy/smolchess-v2 | 1093 |
|
| 60 |
-
| mlabonne/chesspythia-70m | 1088 |
|
| 61 |
-
| **FlameF0X/ChessSLM** | **1087** |
|
| 62 |
| DedeProGames/mini-chennus | 1083 |
|
| 63 |
| distilbert/distilgpt2 | 1061 |
|
| 64 |
-
|
|
| 65 |
| facebook/opt-125m | 1057 |
|
|
|
|
|
|
|
| 66 |
| mlabonne/grandpythia-200k-70m | 1050 |
|
| 67 |
| DedeProGames/Chesser-248K-Mini | 1048 |
|
| 68 |
-
| bharathrajcl/chess_llama_68m | 1046 |
|
| 69 |
|
| 70 |
---
|
| 71 |
|
|
|
|
| 1 |
---
|
| 2 |
license: apache-2.0
|
| 3 |
datasets:
|
| 4 |
+
- mlabonne/chessllm
|
| 5 |
library_name: transformers
|
| 6 |
tags:
|
| 7 |
+
- chess
|
| 8 |
pipeline_tag: text-generation
|
| 9 |
+
new_version: FlameF0X/ChessSLM-RL
|
| 10 |
---
|
| 11 |
|
| 12 |
# ChessSLM
|
|
|
|
| 37 |
ChessSLM can play chess by generating moves sequentially in SAN notation.
|
| 38 |
It has been evaluated in matches against several language models, including:
|
| 39 |
|
| 40 |
+
- Claude [Won against it]
|
| 41 |
+
- Gemini [Lost again it]
|
| 42 |
- Qwen
|
| 43 |
- GPT-2
|
| 44 |
- GPT-Neo
|
|
|
|
| 47 |
- Mistral
|
| 48 |
- other small chess-oriented models
|
| 49 |
|
| 50 |
+
The model achieves an averaging rating of **around ~1054 Elo** against other language models despite its small size.
|
| 51 |
|
| 52 |
---
|
| 53 |
|
|
|
|
| 55 |
|
| 56 |
| Model | Elo Rating |
|
| 57 |
|------|------------|
|
| 58 |
+
| EleutherAI/pythia-70m-deduped | 1111 |
|
| 59 |
+
| mlabonne/chesspythia-70m | 1101 |
|
| 60 |
| nlpguy/amdchess-v9 | 1094 |
|
| 61 |
| nlpguy/smolchess-v2 | 1093 |
|
|
|
|
|
|
|
| 62 |
| DedeProGames/mini-chennus | 1083 |
|
| 63 |
| distilbert/distilgpt2 | 1061 |
|
| 64 |
+
| DedeProGames/dialochess | 1059 |
|
| 65 |
| facebook/opt-125m | 1057 |
|
| 66 |
+
| **FlameF0X/ChessSLM** | **1054** |
|
| 67 |
+
| **FlameF0X/ChessSLM-RL** | **1054** |
|
| 68 |
| mlabonne/grandpythia-200k-70m | 1050 |
|
| 69 |
| DedeProGames/Chesser-248K-Mini | 1048 |
|
|
|
|
| 70 |
|
| 71 |
---
|
| 72 |
|