M31 sft checkpoint 123

experiments/M31-Python-Agent-220M-v5/checkpoints/sft/step-00000123/README.md

@@ -0,0 +1,22 @@
+# M31Tesla — M31-Python-Agent-220M-v5
+
+Independent M31 research checkpoint.
+
+Parameters: 228,953,984
+Vocabulary: 32,768
+Hidden: 896
+Layers: 20
+Attention: 14 Q / 7 KV
+SwiGLU: 2816
+Max context: 2048
+Training sequence: 512
+
+Curriculum:
+1. Python repository pretraining
+2. Python instruction SFT
+3. OpenHands agent trajectories
+4. SWE-smith patch/verification curriculum
+
+HumanEval is evaluation-only.
+
+Dataset licenses remain subject to their respective sources.

experiments/M31-Python-Agent-220M-v5/checkpoints/sft/step-00000123/checkpoint.json

@@ -0,0 +1,10 @@
+{
+  "stage": "sft",
+  "step": 123,
+  "metrics": {
+    "stage_complete": true
+  },
+  "config_hash": "01c040c4d3d046e3f424",
+  "repo_id": "eshanized/M31Tesla",
+  "path": "experiments/M31-Python-Agent-220M-v5/checkpoints/sft/step-00000123"
+}

experiments/M31-Python-Agent-220M-v5/checkpoints/sft/step-00000123/config.json

@@ -0,0 +1,18 @@
+{
+  "model_type": "m31",
+  "architectures": [
+    "M31Model"
+  ],
+  "vocab_size": 32768,
+  "hidden_size": 896,
+  "num_hidden_layers": 20,
+  "num_attention_heads": 14,
+  "num_key_value_heads": 7,
+  "intermediate_size": 2816,
+  "max_position_embeddings": 2048,
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 10000.0,
+  "tie_word_embeddings": true,
+  "parameter_count": 228953984,
+  "experiment": "M31-Python-Agent-220M-v5"
+}

experiments/M31-Python-Agent-220M-v5/checkpoints/sft/step-00000123/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:54c742143c917171f997e1076447a7fb47e5b3f5f2e94fc11ff730bba625da04
+size 915833008