I'm from this: https://www.reddit.com/r/ollama/s/2Jf99CBu2Q
I try your app.
First of all, nice app, but I can't paste long prompt describe what do, I must write from beginning to end. Paste option don't work, I can only copy from app.
App run with llama.cpp and make simple markdown App.
But after minutes I have this error.
From console llama.cpp, this is only info:
W load: control-looking token: 212 '' was not control-type; this is probably a bug in the model. its type will be overridden
0.01.100.657 W load: control-looking token: 50 '<|tool_response>' was not control-type; this is probably a bug in the model. its type will be overridden
0.01.110.004 W load: control-looking token: 1 '' was not control-type; this is probably a bug in the model. its type will be overridden
0.01.137.825 W load: special_eog_ids contains '<|tool_response>', removing '' token from EOG list
0.01.510.668 W warning: failed to mlock 2386145280-byte buffer (after previously locking 0 bytes): Cannot allocate memory
Try increasing RLIMIT_MEMLOCK ('ulimit -l' as root).
0.02.831.086 W llama_context: n_ctx_seq (30208) < n_ctx_train (131072) -- the full capacity of the model will not be utilized
0.02.831.599 W llama_kv_cache: the V embeddings have different sizes across layers and FA is not enabled - padding V cache to 1024
0.02.973.040 W llama_kv_cache: the V embeddings have different sizes across layers and FA is not enabled - padding V cache to 1024
0.03.021.975 W common_init_from_params: warming up the model with an empty run - please wait ... (--no-warmup to disable)
0.03.221.116 I srv load_model: initializing slots, n_slots = 4
0.03.460.793 W common_speculative_init: no implementations specified for speculative decoding
0.03.460.797 I slot load_model: id 0 | task -1 | new slot, n_ctx = 30208
And this brake:
W srv next: stopping wait for next result due to should_stop condition (adjust the --timeout argument if needed)
39.28.895.781 W srv next: ref: ggml-org/llama.cpp#22907
39.28.896.020 W srv stop: cancel task, id_task = 2975
===
✓ bash $ npm init -y 203ms
✓ bash $ npm install express marked 1391ms
✓ write_file ✓ Created data.json (9 lines) 0ms
✓ write_file ✓ Created server.js (132 lines) 1ms
✓ write_file ✓ Created public/index.html (72 lines) 0ms
✓ ✗ fetch failed
✓ ✗ No response from model
If app will have debug mode, then will be simpler to find bug, if we see log from app what step brake.
I'm from this: https://www.reddit.com/r/ollama/s/2Jf99CBu2Q
I try your app.
First of all, nice app, but I can't paste long prompt describe what do, I must write from beginning to end. Paste option don't work, I can only copy from app.
App run with llama.cpp and make simple markdown App.
But after minutes I have this error.
From console llama.cpp, this is only info:
W load: control-looking token: 212 '' was not control-type; this is probably a bug in the model. its type will be overridden
0.01.100.657 W load: control-looking token: 50 '<|tool_response>' was not control-type; this is probably a bug in the model. its type will be overridden
0.01.110.004 W load: control-looking token: 1 '' was not control-type; this is probably a bug in the model. its type will be overridden
0.01.137.825 W load: special_eog_ids contains '<|tool_response>', removing '' token from EOG list
0.01.510.668 W warning: failed to mlock 2386145280-byte buffer (after previously locking 0 bytes): Cannot allocate memory
Try increasing RLIMIT_MEMLOCK ('ulimit -l' as root).
0.02.831.086 W llama_context: n_ctx_seq (30208) < n_ctx_train (131072) -- the full capacity of the model will not be utilized
0.02.831.599 W llama_kv_cache: the V embeddings have different sizes across layers and FA is not enabled - padding V cache to 1024
0.02.973.040 W llama_kv_cache: the V embeddings have different sizes across layers and FA is not enabled - padding V cache to 1024
0.03.021.975 W common_init_from_params: warming up the model with an empty run - please wait ... (--no-warmup to disable)
0.03.221.116 I srv load_model: initializing slots, n_slots = 4
0.03.460.793 W common_speculative_init: no implementations specified for speculative decoding
0.03.460.797 I slot load_model: id 0 | task -1 | new slot, n_ctx = 30208
And this brake:
W srv next: stopping wait for next result due to should_stop condition (adjust the --timeout argument if needed)
39.28.895.781 W srv next: ref: ggml-org/llama.cpp#22907
39.28.896.020 W srv stop: cancel task, id_task = 2975
===
✓ bash $ npm init -y 203ms
✓ bash $ npm install express marked 1391ms
✓ write_file ✓ Created data.json (9 lines) 0ms
✓ write_file ✓ Created server.js (132 lines) 1ms
✓ write_file ✓ Created public/index.html (72 lines) 0ms
✓ ✗ fetch failed
✓ ✗ No response from model
If app will have debug mode, then will be simpler to find bug, if we see log from app what step brake.