Actually clear batch data and make dumping cache use the correct sequ…

…ence ids

common/common.cpp

@@ -1120,6 +1120,15 @@ struct llama_context_params llama_context_params_from_gpt_params(const gpt_param
 }
 
 void llama_batch_clear(struct llama_batch & batch) {
+    for (int token = 0; token < batch.n_tokens; token++) {
+        batch.token[token] = 0;
+        batch.pos[token] = -1;
+        batch.logits[token] = false;
+        for (int seq_id = 0; seq_id < batch.n_seq_id[token]; seq_id++) {
+            batch.seq_id[token][seq_id] = 0;
+        }
+        batch.n_seq_id[token] = 0;
+    }
     batch.n_tokens = 0;
 }
 
@@ -1621,7 +1630,7 @@ void dump_kv_cache_view(const llama_kv_cache_view & view, int row_size) {
         for (int j = 0; j < view.n_max_seq; j++) {
             if (cs_curr[j] >= 0) { seq_count++; }
         }
-        putchar(slot_chars[std::min(sizeof(slot_chars) - 2, size_t(seq_count))]);
+        printf("%4d ",c_curr->pos);
     }
 
     printf("\n=== Done dumping\n");
@@ -1642,7 +1651,7 @@ void dump_kv_cache_view_seqs(const llama_kv_cache_view & view, int row_size) {
             if (cs_curr[j] < 0) { continue; }
             if (seqs.find(cs_curr[j]) == seqs.end()) {
                 if (seqs.size() + 1 >= sizeof(slot_chars)) { break; }
-                seqs[cs_curr[j]] = seqs.size();
+                seqs[cs_curr[j]] = cs_curr[j];
             }
         }
         if (seqs.size() + 1 >= sizeof(slot_chars)) { break; }
@@ -1668,7 +1677,8 @@ void dump_kv_cache_view_seqs(const llama_kv_cache_view & view, int row_size) {
                 putchar('.');
             }
         }
-        putchar(' ');
+        printf(" (%d) ", c_curr->pos);
+        //putchar(' ');
     }
 
     printf("\n=== Done dumping\n");