Jamba-1.5-Large Q2_K GGUF header (mradermacher conversion)

Official ai21labs/AI21-Jamba-Large-1.5 config.json is gated (HTTP 401). This public GGUF converts that checkpoint (general.name is AI21 Jamba Large 1.5). First 4MB of Q2_K part1of3 fetched and GGUF v3 header parsed. Read: block_count 72, embedding_length 8192, feed_forward_length 24576, attention.head_count 64, attention.head_count_kv a 72-long int32 array with 8 at indices 4,12,20,28,36,44,52,60,68 and 0 elsewhere (9 attention layers), ssm.conv_kernel 4, ssm.inner_size 16384, ssm.state_size 16, ssm.time_step_rank 512, context_length 262144, expert_count 16, expert_used_count 2, file_type 10 (Q2_K). head_dim is not stored; embedding_length/head_count = 8192/64 = 128. Third-party conversion, not AI21 own config.json.