Restore 54 KV-shared tensors (layers 24-41) in model.safetensors; stock-schema config.json (root + merged)

README.md

@@ -142,3 +142,29 @@ fixed function-calling instruction block to the caller's system message, and
 when a generation terminates inside the thought channel with an empty answer it
 attempts to recover a bracket-format tool-call list from the raw text before
 returning an empty answer. Both behaviors are in the script, unabridged.
+
+---
+
+### 2026-08 weight-file fix
+
+Earlier revisions of this repository shipped a `model.safetensors` missing 54
+tensors relative to the base model: `k_proj`, `v_proj`, and `k_norm` weights
+for language-model layers 24–41 (2076 keys instead of 2130). These layers use
+shared KV projections (`num_kv_shared_layers: 18`), and the export path
+(transformers 5.15.0 `save_pretrained`) silently omitted their parameters.
+Some engines tolerated the omission; vLLM 0.25.1 correctly refused to load
+the checkpoint.
+
+The current revision restores all 2130 tensors. The 54 restored tensors are
+byte-exact copies from `google/gemma-4-E4B-it` — this is provably lossless:
+the fine-tuning adapter contains no LoRA entries for k/v projections at or
+above layer 24, and every one of the 2076 retained tensors is either
+bit-identical to the base or accounted for by an adapter-targeted module.
+`config.json` now matches the base model's schema so the checkpoint loads
+under both current and older transformers versions.
+
+Verification: the fixed checkpoint serves under vLLM 0.25.1, and greedy
+decoding logprobs match an independently reconstructed base+adapter merge at
+the same numerical floor measured between two servings of the unmodified base
+model (top-1 agreement 99.6% vs a 98.8% same-model cross-engine control; max
+chosen-token logprob difference 0.166 vs 0.126 control).

config.json

@@ -66,6 +66,7 @@
     "eos_token_id": 1,
     "expert_intermediate_size": null,
     "final_logit_softcapping": 30.0,
+    "global_head_dim": 512,
     "head_dim": 256,
     "hidden_activation": "gelu_pytorch_tanh",
     "hidden_size": 2560,
@@ -118,36 +119,13 @@
     ],
     "max_position_embeddings": 131072,
     "model_type": "gemma4_text",
-    "moe_intermediate_size": null,
     "num_attention_heads": 8,
     "num_experts": null,
+    "num_global_key_value_heads": null,
     "num_hidden_layers": 42,
     "num_key_value_heads": 2,
     "num_kv_shared_layers": 18,
     "pad_token_id": 0,
-    "per_layer_config": {
-      "05": {
-        "head_dim": 512
-      },
-      "11": {
-        "head_dim": 512
-      },
-      "17": {
-        "head_dim": 512
-      },
-      "23": {
-        "head_dim": 512
-      },
-      "29": {
-        "head_dim": 512
-      },
-      "35": {
-        "head_dim": 512
-      },
-      "41": {
-        "head_dim": 512
-      }
-    },
     "rms_norm_eps": 1e-06,
     "rope_parameters": {
       "full_attention": {
@@ -170,7 +148,7 @@
     "vocab_size_per_layer_input": 262144
   },
   "tie_word_embeddings": true,
-  "transformers_version": "5.15.0",
+  "transformers_version": "5.5.0.dev0",
   "video_token_id": 258884,
   "vision_config": {
     "_name_or_path": "",

merged/config.json

@@ -66,6 +66,7 @@
     "eos_token_id": 1,
     "expert_intermediate_size": null,
     "final_logit_softcapping": 30.0,
+    "global_head_dim": 512,
     "head_dim": 256,
     "hidden_activation": "gelu_pytorch_tanh",
     "hidden_size": 2560,
@@ -118,36 +119,13 @@
     ],
     "max_position_embeddings": 131072,
     "model_type": "gemma4_text",
-    "moe_intermediate_size": null,
     "num_attention_heads": 8,
     "num_experts": null,
+    "num_global_key_value_heads": null,
     "num_hidden_layers": 42,
     "num_key_value_heads": 2,
     "num_kv_shared_layers": 18,
     "pad_token_id": 0,
-    "per_layer_config": {
-      "05": {
-        "head_dim": 512
-      },
-      "11": {
-        "head_dim": 512
-      },
-      "17": {
-        "head_dim": 512
-      },
-      "23": {
-        "head_dim": 512
-      },
-      "29": {
-        "head_dim": 512
-      },
-      "35": {
-        "head_dim": 512
-      },
-      "41": {
-        "head_dim": 512
-      }
-    },
     "rms_norm_eps": 1e-06,
     "rope_parameters": {
       "full_attention": {
@@ -170,7 +148,7 @@
     "vocab_size_per_layer_input": 262144
   },
   "tie_word_embeddings": true,
-  "transformers_version": "5.15.0",
+  "transformers_version": "5.5.0.dev0",
   "video_token_id": 258884,
   "vision_config": {
     "_name_or_path": "",

merged/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4399a604e0f72d9785060b3dcce47fad44986da0e8dfd82ef17db116dad9ba51
-size 15882477500
+oid sha256:85ba6e826cae2cb4978659417dda0840e738a6adc0bba8a7dfd9b43499947faf
+size 15992595884

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4399a604e0f72d9785060b3dcce47fad44986da0e8dfd82ef17db116dad9ba51
-size 15882477500
+oid sha256:85ba6e826cae2cb4978659417dda0840e738a6adc0bba8a7dfd9b43499947faf
+size 15992595884