Update README.md
Browse files
README.md
CHANGED
|
@@ -103,13 +103,13 @@ The model was evaluated on AIME24, GPQA Diamond, and MATH-500 benchmarks using t
|
|
| 103 |
The results were obtained using the following commands, with custom tasks and 10 rounds using different random seeds.
|
| 104 |
|
| 105 |
```
|
| 106 |
-
MODEL_ARGS="model_name=amd/DeepSeek-R1-0528-MXFP4-
|
| 107 |
-
OUTPUT_DIR="results/DeepSeek-R1-0528-MXFP4-
|
| 108 |
LOG="logs/deepseek_0528_maxfp4.log"
|
| 109 |
|
| 110 |
lighteval vllm $MODEL_ARGS "custom|aime24_single|0|0,custom|math_500_single|0|0,custom|gpqa:diamond_single|0|0" \
|
| 111 |
--use-chat-template \
|
| 112 |
-
--output-dir "
|
| 113 |
2>&1 | tee -a "$LOG"
|
| 114 |
```
|
| 115 |
|
|
|
|
| 103 |
The results were obtained using the following commands, with custom tasks and 10 rounds using different random seeds.
|
| 104 |
|
| 105 |
```
|
| 106 |
+
MODEL_ARGS="model_name=amd/DeepSeek-R1-0528-MXFP4-Preview,dtype=bfloat16,tensor_parallel_size=8,max_model_length=71536,max_num_batched_tokens=32768,gpu_memory_utilization=0.85,generation_parameters={max_new_tokens:65536,temperature:0.6,top_p:0.95,seed:$SEED}"
|
| 107 |
+
OUTPUT_DIR="results/DeepSeek-R1-0528-MXFP4-Preview-Seed"
|
| 108 |
LOG="logs/deepseek_0528_maxfp4.log"
|
| 109 |
|
| 110 |
lighteval vllm $MODEL_ARGS "custom|aime24_single|0|0,custom|math_500_single|0|0,custom|gpqa:diamond_single|0|0" \
|
| 111 |
--use-chat-template \
|
| 112 |
+
--output-dir "$OUTPUT_DIR/seed_$SEED" \
|
| 113 |
2>&1 | tee -a "$LOG"
|
| 114 |
```
|
| 115 |
|