Instructions to use LLM-OS-Models/KoHRM-Text-1.4B-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- llama.cpp
How to use LLM-OS-Models/KoHRM-Text-1.4B-GGUF with llama.cpp:
Install (macOS, Linux)
curl -LsSf https://llama.app/install.sh | sh # Start a local OpenAI-compatible server with a web UI: llama serve -hf LLM-OS-Models/KoHRM-Text-1.4B-GGUF:Q4_K_M # Run inference directly in the terminal: llama cli -hf LLM-OS-Models/KoHRM-Text-1.4B-GGUF:Q4_K_M
Install from WinGet (Windows)
winget install llama.cpp # Start a local OpenAI-compatible server with a web UI: llama serve -hf LLM-OS-Models/KoHRM-Text-1.4B-GGUF:Q4_K_M # Run inference directly in the terminal: llama cli -hf LLM-OS-Models/KoHRM-Text-1.4B-GGUF:Q4_K_M
Use pre-built binary
# Download pre-built binary from: # https://github.com/ggerganov/llama.cpp/releases # Start a local OpenAI-compatible server with a web UI: ./llama-server -hf LLM-OS-Models/KoHRM-Text-1.4B-GGUF:Q4_K_M # Run inference directly in the terminal: ./llama-cli -hf LLM-OS-Models/KoHRM-Text-1.4B-GGUF:Q4_K_M
Build from source code
git clone https://github.com/ggerganov/llama.cpp.git cd llama.cpp cmake -B build cmake --build build -j --target llama-server llama-cli # Start a local OpenAI-compatible server with a web UI: ./build/bin/llama-server -hf LLM-OS-Models/KoHRM-Text-1.4B-GGUF:Q4_K_M # Run inference directly in the terminal: ./build/bin/llama-cli -hf LLM-OS-Models/KoHRM-Text-1.4B-GGUF:Q4_K_M
Use Docker
docker model run hf.co/LLM-OS-Models/KoHRM-Text-1.4B-GGUF:Q4_K_M
- LM Studio
- Jan
- vLLM
How to use LLM-OS-Models/KoHRM-Text-1.4B-GGUF with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "LLM-OS-Models/KoHRM-Text-1.4B-GGUF" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "LLM-OS-Models/KoHRM-Text-1.4B-GGUF", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/LLM-OS-Models/KoHRM-Text-1.4B-GGUF:Q4_K_M
- Ollama
How to use LLM-OS-Models/KoHRM-Text-1.4B-GGUF with Ollama:
ollama run hf.co/LLM-OS-Models/KoHRM-Text-1.4B-GGUF:Q4_K_M
- Unsloth Studio
How to use LLM-OS-Models/KoHRM-Text-1.4B-GGUF with Unsloth Studio:
Install Unsloth Studio (macOS, Linux, WSL)
curl -fsSL https://unsloth.ai/install.sh | sh # Run unsloth studio unsloth studio -H 0.0.0.0 -p 8888 # Then open http://localhost:8888 in your browser # Search for LLM-OS-Models/KoHRM-Text-1.4B-GGUF to start chatting
Install Unsloth Studio (Windows)
irm https://unsloth.ai/install.ps1 | iex # Run unsloth studio unsloth studio -H 0.0.0.0 -p 8888 # Then open http://localhost:8888 in your browser # Search for LLM-OS-Models/KoHRM-Text-1.4B-GGUF to start chatting
Using HuggingFace Spaces for Unsloth
# No setup required # Open https://huggingface.co/spaces/unsloth/studio in your browser # Search for LLM-OS-Models/KoHRM-Text-1.4B-GGUF to start chatting
- Docker Model Runner
How to use LLM-OS-Models/KoHRM-Text-1.4B-GGUF with Docker Model Runner:
docker model run hf.co/LLM-OS-Models/KoHRM-Text-1.4B-GGUF:Q4_K_M
- Lemonade
How to use LLM-OS-Models/KoHRM-Text-1.4B-GGUF with Lemonade:
Pull the model
# Download Lemonade from https://lemonade-server.ai/ lemonade pull LLM-OS-Models/KoHRM-Text-1.4B-GGUF:Q4_K_M
Run and chat with the model
lemonade run user.KoHRM-Text-1.4B-GGUF-Q4_K_M
List all available models
lemonade list
- Atomic Chat
Update model card with representative finance QA smoke test
Browse files- README.md +35 -22
- reports/conversion_report.md +8 -3
- reports/generation_quality_check.md +55 -21
- reports/generation_tests/finance_short_q8_0.txt +46 -0
README.md
CHANGED
|
@@ -62,7 +62,7 @@ Download a quantized GGUF file:
|
|
| 62 |
|
| 63 |
```bash
|
| 64 |
huggingface-cli download LLM-OS-Models/KoHRM-Text-1.4B-GGUF \
|
| 65 |
-
KoHRM-Text-1.4B-
|
| 66 |
--local-dir .
|
| 67 |
```
|
| 68 |
|
|
@@ -70,57 +70,70 @@ Run on CPU:
|
|
| 70 |
|
| 71 |
```bash
|
| 72 |
./build-hrm/bin/llama-cli \
|
| 73 |
-
-m ./KoHRM-Text-1.4B-
|
| 74 |
-ngl 0 \
|
| 75 |
-t 4 \
|
| 76 |
-
-c
|
| 77 |
-
-n
|
| 78 |
-
--seed
|
| 79 |
-
--temp 0.
|
| 80 |
--top-p 0.9 \
|
|
|
|
| 81 |
--single-turn \
|
| 82 |
--simple-io \
|
| 83 |
--no-warmup \
|
| 84 |
--display-prompt \
|
| 85 |
-
-p $'
|
| 86 |
```
|
| 87 |
|
| 88 |
## CPU Generation Tests
|
| 89 |
|
| 90 |
Tested locally on CPU with the patched `llama.cpp` build and the prompt shown above.
|
| 91 |
|
| 92 |
-
|
| 93 |
-
|
| 94 |
-
| `KoHRM-Text-1.4B-Q4_K_M.gguf` | 33.6 t/s | 13.2 t/s | Runtime OK, not suitable as a legal-answer quality example |
|
| 95 |
-
| `KoHRM-Text-1.4B-Q8_0.gguf` | 23.8 t/s | 6.8 t/s | Runtime OK, not suitable as a legal-answer quality example |
|
| 96 |
-
|
| 97 |
-
Q4_K_M output excerpt:
|
| 98 |
|
| 99 |
```text
|
| 100 |
-
|
| 101 |
-
๋ฏธ์ฑ๋
์๋ ๋ฏธ์ฑ๋
์๋ฅผ ๋์์ผ๋ก ํ๋ ๋ฒ๋ฅ ํ์ ... ๋ฏผ๋ฒ ์ 999-1001 ...
|
| 102 |
```
|
| 103 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 104 |
Q8_0 output excerpt:
|
| 105 |
|
| 106 |
```text
|
| 107 |
-
|
| 108 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 109 |
```
|
| 110 |
|
| 111 |
-
|
| 112 |
-
|
| 113 |
-
The full smoke-test logs are in `reports/`.
|
| 114 |
|
| 115 |
## Prompt Format
|
| 116 |
|
| 117 |
-
The source
|
| 118 |
|
| 119 |
```text
|
| 120 |
<|im_start|><|object_ref_start|>PROMPT<|im_end|>
|
| 121 |
```
|
| 122 |
|
| 123 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 124 |
|
| 125 |
## GGUF Metadata
|
| 126 |
|
|
|
|
| 62 |
|
| 63 |
```bash
|
| 64 |
huggingface-cli download LLM-OS-Models/KoHRM-Text-1.4B-GGUF \
|
| 65 |
+
KoHRM-Text-1.4B-Q8_0.gguf \
|
| 66 |
--local-dir .
|
| 67 |
```
|
| 68 |
|
|
|
|
| 70 |
|
| 71 |
```bash
|
| 72 |
./build-hrm/bin/llama-cli \
|
| 73 |
+
-m ./KoHRM-Text-1.4B-Q8_0.gguf \
|
| 74 |
-ngl 0 \
|
| 75 |
-t 4 \
|
| 76 |
+
-c 1024 \
|
| 77 |
+
-n 260 \
|
| 78 |
+
--seed 41 \
|
| 79 |
+
--temp 0.45 \
|
| 80 |
--top-p 0.9 \
|
| 81 |
+
--repeat-penalty 1.08 \
|
| 82 |
--single-turn \
|
| 83 |
--simple-io \
|
| 84 |
--no-warmup \
|
| 85 |
--display-prompt \
|
| 86 |
+
-p $'ํด์ธ์ฃผ์ ํฌ์์์ ์/๋ฌ๋ฌ ํ์จ ๋ณ๋์ด ์ํ ์์ต๋ฅ ์ ๋ฏธ์น๋ ์ํฅ๊ณผ ๋์ ๋ฐฉ์์ ๊ฐ๋จํ ์ค๋ช
ํด ์ฃผ์ธ์.'
|
| 87 |
```
|
| 88 |
|
| 89 |
## CPU Generation Tests
|
| 90 |
|
| 91 |
Tested locally on CPU with the patched `llama.cpp` build and the prompt shown above.
|
| 92 |
|
| 93 |
+
This prompt was chosen after checking the KoHRM training-data path. KoHRM uses
|
| 94 |
+
the HRM `V1Dataset` instruction-response layout:
|
|
|
|
|
|
|
|
|
|
|
|
|
| 95 |
|
| 96 |
```text
|
| 97 |
+
<|im_start|><condition_token>instruction<|im_end|>response<|box_end|>
|
|
|
|
| 98 |
```
|
| 99 |
|
| 100 |
+
The instruction/prefix span is not trained with loss; the response span is
|
| 101 |
+
trained with response-only loss. Local decoded samples include short Korean
|
| 102 |
+
finance QA rows, so the representative GGUF smoke prompt below uses the same
|
| 103 |
+
plain instruction style instead of a legal reasoning prompt.
|
| 104 |
+
|
| 105 |
+
| file | prompt speed | generation speed | value check |
|
| 106 |
+
|---|---:|---:|---|
|
| 107 |
+
| `KoHRM-Text-1.4B-Q8_0.gguf` | 25.3 t/s | 5.0 t/s | Runtime OK; useful qualitative finance QA sample |
|
| 108 |
+
|
| 109 |
Q8_0 output excerpt:
|
| 110 |
|
| 111 |
```text
|
| 112 |
+
1. **ํ์จ ๋ณ๋์ ์ํฅ:** ํด์ธ ์ฃผ์ ํฌ์์ ์์ต๋ฅ ์ ์ฃผ๊ฐ ์์น์ ๋ฐ๋ฅธ ์์ต๋ฟ๋ง ์๋๋ผ ํ์จ ๋ณ๋์ ๋ฐ๋ฅธ ์์ต ๋๋ ์์ค์ ํฌํจํฉ๋๋ค.
|
| 113 |
+
...
|
| 114 |
+
2. **๋์ ๋ฐฉ์:**
|
| 115 |
+
- **ํ์จ ๋ณ๋ ์ํ ๊ด๋ฆฌ:** ํ์จ ๋ณ๋ ์ํ์ ์ค์ด๊ธฐ ์ํด ํํค์ง ์ํ์ ํ์ฉํ๊ฑฐ๋, ๋ถํ ๋งค์/๋งค๋ ์ ๋ต์ ํตํด ํ์จ ๋ณ๋์ ๋ฐ๋ฅธ ์ํฅ์ ์ํํ ์ ์์ต๋๋ค.
|
| 116 |
+
- **์ฅ๊ธฐ ํฌ์:** ์ฅ๊ธฐ ํฌ์๋ฅผ ํตํด ํ์จ ๋ณ๋์ ๋จ๊ธฐ์ ์ธ ์ํฅ์ ์ํํ๊ณ , ์ฅ๊ธฐ์ ์ธ ์ฃผ๊ฐ ์์น์ ์ง์คํ ์ ์์ต๋๋ค.
|
| 117 |
+
- **๋ถ์ฐ ํฌ์:** ๋ค์ํ ๊ตญ๊ฐ์ ์ฃผ์์ ๋ถ์ฐ ํฌ์ํ์ฌ ํน์ ๊ตญ๊ฐ์ ํ์จ ๋ณ๋ ์ํ์ ์ค์ผ ์ ์์ต๋๋ค.
|
| 118 |
```
|
| 119 |
|
| 120 |
+
The full smoke-test log is in `reports/generation_tests/finance_short_q8_0.txt`.
|
| 121 |
+
This is a qualitative CPU runtime sample, not a benchmark or financial advice.
|
|
|
|
| 122 |
|
| 123 |
## Prompt Format
|
| 124 |
|
| 125 |
+
The source training/inference wrapper is:
|
| 126 |
|
| 127 |
```text
|
| 128 |
<|im_start|><|object_ref_start|>PROMPT<|im_end|>
|
| 129 |
```
|
| 130 |
|
| 131 |
+
`prepare_sft_data.py` writes the generic HRM V1Dataset layout with
|
| 132 |
+
`direct=<|object_ref_start|>` by default. In this patched GGUF runtime,
|
| 133 |
+
`llama-completion` could load the model but returned an immediate end token for
|
| 134 |
+
the tested prompts, while `llama-cli --single-turn` produced visible CPU token
|
| 135 |
+
generation. The public checkpoint is a rolling pretraining-stage model, not a
|
| 136 |
+
final chat/SFT model, so instruction following can still be unstable.
|
| 137 |
|
| 138 |
## GGUF Metadata
|
| 139 |
|
reports/conversion_report.md
CHANGED
|
@@ -93,11 +93,16 @@ Exiting...
|
|
| 93 |
|
| 94 |
The generated text is not a legal-quality answer; this source checkpoint is documented as a pretraining-stage model, not an instruction-tuned chat model.
|
| 95 |
|
| 96 |
-
Additional CPU generation tests were run on 2026-06-14
|
|
|
|
|
|
|
|
|
|
| 97 |
|
| 98 |
```text
|
| 99 |
-
|
| 100 |
-
|
|
|
|
|
|
|
| 101 |
```
|
| 102 |
|
| 103 |
## Artifacts
|
|
|
|
| 93 |
|
| 94 |
The generated text is not a legal-quality answer; this source checkpoint is documented as a pretraining-stage model, not an instruction-tuned chat model.
|
| 95 |
|
| 96 |
+
Additional CPU generation tests were run on 2026-06-14. After checking the KoHRM
|
| 97 |
+
training-data layout, the public model-card example was changed from a legal QA
|
| 98 |
+
prompt to a short Korean finance QA prompt that better matches the local decoded
|
| 99 |
+
prepared samples and the instruction-response preprocessing path.
|
| 100 |
|
| 101 |
```text
|
| 102 |
+
Q8_0 prompt:
|
| 103 |
+
ํด์ธ์ฃผ์ ํฌ์์์ ์/๋ฌ๋ฌ ํ์จ ๋ณ๋์ด ์ํ ์์ต๋ฅ ์ ๋ฏธ์น๋ ์ํฅ๊ณผ ๋์ ๋ฐฉ์์ ๊ฐ๋จํ ์ค๋ช
ํด ์ฃผ์ธ์.
|
| 104 |
+
|
| 105 |
+
Q8_0: Prompt 25.3 t/s, Generation 5.0 t/s, runtime OK, useful qualitative finance QA sample
|
| 106 |
```
|
| 107 |
|
| 108 |
## Artifacts
|
reports/generation_quality_check.md
CHANGED
|
@@ -2,48 +2,82 @@
|
|
| 2 |
|
| 3 |
Date: 2026-06-14 UTC
|
| 4 |
|
| 5 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 6 |
|
| 7 |
```text
|
| 8 |
-
|
| 9 |
-
๋ต๋ณ:
|
| 10 |
```
|
| 11 |
|
| 12 |
-
Command
|
| 13 |
|
| 14 |
```bash
|
| 15 |
./build-hrm/bin/llama-cli \
|
| 16 |
-
-m
|
| 17 |
-ngl 0 \
|
| 18 |
-t 4 \
|
| 19 |
-
-c
|
| 20 |
-
-n
|
| 21 |
-
--seed
|
| 22 |
-
--temp 0.
|
| 23 |
--top-p 0.9 \
|
|
|
|
| 24 |
--single-turn \
|
| 25 |
--simple-io \
|
| 26 |
--no-warmup \
|
| 27 |
--display-prompt \
|
| 28 |
-
-p $'
|
| 29 |
```
|
| 30 |
|
| 31 |
-
##
|
| 32 |
|
| 33 |
| model | prompt speed | generation speed | verdict |
|
| 34 |
|---|---:|---:|---|
|
| 35 |
-
| `KoHRM-Text-1.4B-
|
| 36 |
-
|
|
|
|
| 37 |
|
| 38 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 39 |
|
| 40 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 41 |
|
| 42 |
-
|
| 43 |
|
| 44 |
-
-
|
| 45 |
-
|
| 46 |
-
|
| 47 |
-
- The Q8_0 sample is semantically incoherent around `์์ '๋ฏธ์ฑ๋
์'`, `์ฒ๋ถ`, and `์๋ฌด`.
|
| 48 |
|
| 49 |
-
|
|
|
|
|
|
| 2 |
|
| 3 |
Date: 2026-06-14 UTC
|
| 4 |
|
| 5 |
+
## Data Format Check
|
| 6 |
+
|
| 7 |
+
Before selecting the representative generation example, the local KoHRM data
|
| 8 |
+
pipeline was checked from the repository docs and code.
|
| 9 |
+
|
| 10 |
+
KoHRM uses the HRM `V1Dataset` instruction-response layout:
|
| 11 |
+
|
| 12 |
+
```text
|
| 13 |
+
<|im_start|><condition_token>instruction<|im_end|>response<|box_end|>
|
| 14 |
+
```
|
| 15 |
+
|
| 16 |
+
Important preprocessing details:
|
| 17 |
+
|
| 18 |
+
- `scripts/prepare_sft_data.py` accepts rows with `instruction`, `response`, and
|
| 19 |
+
optional `condition`.
|
| 20 |
+
- The default direct condition is `direct=<|object_ref_start|>`.
|
| 21 |
+
- `dataset_new.py` reads instruction and response spans separately.
|
| 22 |
+
- With `target_only=True`, the instruction/prefix span has ignore labels and
|
| 23 |
+
the response span receives the supervised loss.
|
| 24 |
+
- Local decoded prepared samples include short Korean finance QA examples, so a
|
| 25 |
+
finance QA prompt is a better smoke-test example than a legal reasoning prompt.
|
| 26 |
+
|
| 27 |
+
## Representative Prompt
|
| 28 |
|
| 29 |
```text
|
| 30 |
+
ํด์ธ์ฃผ์ ํฌ์์์ ์/๋ฌ๋ฌ ํ์จ ๋ณ๋์ด ์ํ ์์ต๋ฅ ์ ๋ฏธ์น๋ ์ํฅ๊ณผ ๋์ ๋ฐฉ์์ ๊ฐ๋จํ ์ค๋ช
ํด ์ฃผ์ธ์.
|
|
|
|
| 31 |
```
|
| 32 |
|
| 33 |
+
Command:
|
| 34 |
|
| 35 |
```bash
|
| 36 |
./build-hrm/bin/llama-cli \
|
| 37 |
+
-m KoHRM-Text-1.4B-Q8_0.gguf \
|
| 38 |
-ngl 0 \
|
| 39 |
-t 4 \
|
| 40 |
+
-c 1024 \
|
| 41 |
+
-n 260 \
|
| 42 |
+
--seed 41 \
|
| 43 |
+
--temp 0.45 \
|
| 44 |
--top-p 0.9 \
|
| 45 |
+
--repeat-penalty 1.08 \
|
| 46 |
--single-turn \
|
| 47 |
--simple-io \
|
| 48 |
--no-warmup \
|
| 49 |
--display-prompt \
|
| 50 |
+
-p $'ํด์ธ์ฃผ์ ํฌ์์์ ์/๋ฌ๋ฌ ํ์จ ๋ณ๋์ด ์ํ ์์ต๋ฅ ์ ๋ฏธ์น๋ ์ํฅ๊ณผ ๋์ ๋ฐฉ์์ ๊ฐ๋จํ ์ค๋ช
ํด ์ฃผ์ธ์.'
|
| 51 |
```
|
| 52 |
|
| 53 |
+
## Result
|
| 54 |
|
| 55 |
| model | prompt speed | generation speed | verdict |
|
| 56 |
|---|---:|---:|---|
|
| 57 |
+
| `KoHRM-Text-1.4B-Q8_0.gguf` | 25.3 t/s | 5.0 t/s | Runtime OK; useful qualitative finance QA sample |
|
| 58 |
+
|
| 59 |
+
Excerpt:
|
| 60 |
|
| 61 |
+
```text
|
| 62 |
+
1. **ํ์จ ๋ณ๋์ ์ํฅ:** ํด์ธ ์ฃผ์ ํฌ์์ ์์ต๋ฅ ์ ์ฃผ๊ฐ ์์น์ ๋ฐ๋ฅธ ์์ต๋ฟ๋ง ์๋๋ผ ํ์จ ๋ณ๋์ ๋ฐ๋ฅธ ์์ต ๋๋ ์์ค์ ํฌํจํฉ๋๋ค.
|
| 63 |
+
...
|
| 64 |
+
2. **๋์ ๋ฐฉ์:**
|
| 65 |
+
- **ํ์จ ๋ณ๋ ์ํ ๊ด๋ฆฌ:** ํ์จ ๋ณ๋ ์ํ์ ์ค์ด๊ธฐ ์ํด ํํค์ง ์ํ์ ํ์ฉํ๊ฑฐ๋, ๋ถํ ๋งค์/๋งค๋ ์ ๋ต์ ํตํด ํ์จ ๋ณ๋์ ๋ฐ๋ฅธ ์ํฅ์ ์ํํ ์ ์์ต๋๋ค.
|
| 66 |
+
- **์ฅ๊ธฐ ํฌ์:** ์ฅ๊ธฐ ํฌ์๋ฅผ ํตํด ํ์จ ๋ณ๋์ ๋จ๊ธฐ์ ์ธ ์ํฅ์ ์ํํ๊ณ , ์ฅ๊ธฐ์ ์ธ ์ฃผ๊ฐ ์์น์ ์ง์คํ ์ ์์ต๋๋ค.
|
| 67 |
+
- **๋ถ์ฐ ํฌ์:** ๋ค์ํ ๊ตญ๊ฐ์ ์ฃผ์์ ๋ถ์ฐ ํฌ์ํ์ฌ ํน์ ๊ตญ๊ฐ์ ํ์จ ๋ณ๋ ์ํ์ ์ค์ผ ์ ์์ต๋๋ค.
|
| 68 |
+
```
|
| 69 |
|
| 70 |
+
Full log:
|
| 71 |
+
|
| 72 |
+
```text
|
| 73 |
+
reports/generation_tests/finance_short_q8_0.txt
|
| 74 |
+
```
|
| 75 |
|
| 76 |
+
## Notes
|
| 77 |
|
| 78 |
+
`llama-completion` loaded the patched GGUF but returned an immediate end token
|
| 79 |
+
for tested prompts. `llama-cli --single-turn` produced visible CPU token
|
| 80 |
+
generation and is therefore used for the model-card smoke command.
|
|
|
|
| 81 |
|
| 82 |
+
This is a qualitative CPU runtime check, not a benchmark, not financial advice,
|
| 83 |
+
and not evidence that the pretraining-stage checkpoint is a final chat/SFT model.
|
reports/generation_tests/finance_short_q8_0.txt
ADDED
|
@@ -0,0 +1,46 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
warning: no usable GPU found, --gpu-layers option will be ignored
|
| 2 |
+
warning: one possible reason is that llama.cpp was compiled without GPU support
|
| 3 |
+
warning: consult docs/build.md for compilation instructions
|
| 4 |
+
|
| 5 |
+
Loading model...
|
| 6 |
+
|
| 7 |
+
|
| 8 |
+
โโ โโ
|
| 9 |
+
โโ โโ
|
| 10 |
+
โโ โโ โโโโ โโโโโโโโ โโโโ โโโโโ โโโโโ โโโโโ
|
| 11 |
+
โโ โโ โโโโโ โโ โโ โโ โโโโโ โโ โโ โโ โโ โโ
|
| 12 |
+
โโ โโ โโโโโ โโ โโ โโ โโโโโ โโ โโโโโ โโโโโ โโโโโ
|
| 13 |
+
โโ โโ
|
| 14 |
+
โโ โโ
|
| 15 |
+
|
| 16 |
+
build : b9263-6a257d446
|
| 17 |
+
model : KoHRM-Text-1.4B-Q8_0.gguf
|
| 18 |
+
modalities : text
|
| 19 |
+
|
| 20 |
+
available commands:
|
| 21 |
+
/exit or Ctrl+C stop or exit
|
| 22 |
+
/regen regenerate the last response
|
| 23 |
+
/clear clear the chat history
|
| 24 |
+
/read <file> add a text file
|
| 25 |
+
/glob <pattern> add text files using globbing pattern
|
| 26 |
+
|
| 27 |
+
|
| 28 |
+
> ํด์ธ์ฃผ์ ํฌ์์์ ์/๋ฌ๋ฌ ํ์จ ๋ณ๋์ด ์ํ ์์ต๋ฅ ์ ๋ฏธ์น๋ ์ํฅ๊ณผ ๋์ ๋ฐฉ์์ ๊ฐ๋จํ ์ค๋ช
ํด ์ฃผ์ธ์.
|
| 29 |
+
|
| 30 |
+
|
| 31 |
+
1. **ํ์จ ๋ณ๋์ ์ํฅ:** ํด์ธ ์ฃผ์ ํฌ์์ ์์ต๋ฅ ์ ์ฃผ๊ฐ ์์น์ ๋ฐ๋ฅธ ์์ต๋ฟ๋ง ์๋๋ผ ํ์จ ๋ณ๋์ ๋ฐ๋ฅธ ์์ต ๋๋ ์์ค์ ํฌํจํฉ๋๋ค. ์๋ฅผ ๋ค์ด, ๋ฏธ๊ตญ ์ฃผ์์ ๋งค์ํ์ ๋ ํ์จ์ด 1,200์์ด์๊ณ , ๋งค๋ ์์ ์ ํ์จ์ด 1,300์์ผ๋ก ์์นํ๋ค๋ฉด, ์ฃผ๊ฐ ์์น๊ณผ ํจ๊ป ํ์ฐจ์ต์ ์ป์ ์ ์์ต๋๋ค. ๋ฐ๋๋ก, ํ์จ์ด ํ๋ฝํ๋ค๋ฉด ํ์ฐจ์์ด ๋ฐ์ํ์ฌ ์ ์ฒด ์์ต๋ฅ ์ด ๋ฎ์์ง ์ ์์ต๋๋ค.
|
| 32 |
+
|
| 33 |
+
2. **๋์ ๋ฐฉ์:**
|
| 34 |
+
- **ํ์จ ๋ณ๋ ์ํ ๊ด๋ฆฌ:** ํ์จ ๋ณ๋ ์ํ์ ์ค์ด๊ธฐ ์ํด ํํค์ง ์ํ์ ํ์ฉํ๊ฑฐ๋, ๋ถํ ๋งค์/๋งค๋ ์ ๋ต์ ํตํด ํ์จ ๋ณ๋์ ๋ฐ๋ฅธ ์ํฅ์ ์ํํ ์ ์์ต๋๋ค.
|
| 35 |
+
- **์ฅ๊ธฐ ํฌ์:** ์ฅ๊ธฐ ํฌ์๋ฅผ ํตํด ํ์จ ๋ณ๋์ ๋จ๊ธฐ์ ์ธ ์ํฅ์ ์ํํ๊ณ , ์ฅ๊ธฐ์ ์ธ ์ฃผ๊ฐ ์์น์ ์ง์คํ ์ ์์ต๋๋ค.
|
| 36 |
+
- **๋ถ์ฐ ํฌ์:** ๋ค์ํ ๊ตญ๊ฐ์ ์ฃผ์์ ๋ถ์ฐ ํฌ์ํ์ฌ ํน์ ๊ตญ๊ฐ์ ํ์จ ๋ณ๋ ์ํ์ ์ค์ผ ์ ์์ต๋๋ค.
|
| 37 |
+
- **ํ์จ ์ ๋ง:** ํ์จ ์ ๋ง์ ์ฐธ๊ณ ํ์ฌ ํฌ์ ์์ ์ ์กฐ์ ํ๊ฑฐ๋, ํ์จ ๋ณ๋์ ๋ฐ๋ฅธ ์์ค์ ์ต์ํํ๊ธฐ ์ํ ์ ๋ต์ ์๋ฆฝํ ์ ์์ต๋๋ค.
|
| 38 |
+
|
| 39 |
+
3. **์ถ๊ฐ ๊ณ ๋ ค ์ฌํญ:**
|
| 40 |
+
- **ํฌ์ ๋ชฉํ:** ํฌ์ ๋ชฉํ์ ๋ฐ๋ผ ์ ์ ํ ๋์ ๋ฐฉ์์ ์ ํํด์ผ ํฉ๋๋ค.
|
| 41 |
+
- **์ํ ๊ฐ์ ์์ค:** ์ํ ๊ฐ์ ์์ค์ ๋ฐ๋ผ ํ์จ ๋ณ๋ ์ํ์ ๊ด๋ฆฌํ ์ ์๋ ์ ๋๋ฅผ ๊ฒฐ์ ํด์ผ ํฉ๋๋ค.
|
| 42 |
+
- **์ ๋ฌธ๊ฐ ์๋ฌธ:** ํ์์ ๋ฐ๋ผ ๊ธ์ต ์ ๋ฌธ๊ฐ์ ์๋ฌธ์ ๋ฐ์
|
| 43 |
+
|
| 44 |
+
[ Prompt: 25.3 t/s | Generation: 5.0 t/s ]
|
| 45 |
+
|
| 46 |
+
Exiting...
|